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 1.36 by spiga, Thu Nov 19 18:04:20 2009 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	>	import os
12	>
13	>
14	>
15	>	class DBSError(exceptions.Exception):
16	>	def __init__(self, errorName, errorMessage):
17	>	args='\nERROR DBS %s : %s \n'%(errorName,errorMessage)
18	>	exceptions.Exception.__init__(self, args)
19	>	pass
20	>
21	>	def getErrorMessage(self):
22	>	""" Return error message """
23	>	return "%s" % (self.args)
24	>
25	>
26	>
27	>	class DBSInvalidDataTierError(exceptions.Exception):
28	>	def __init__(self, errorName, errorMessage):
29	>	args='\nERROR DBS %s : %s \n'%(errorName,errorMessage)
30	>	exceptions.Exception.__init__(self, args)
31	>	pass
32	>
33	>	def getErrorMessage(self):
34	>	""" Return error message """
35	>	return "%s" % (self.args)
36	>
37	>
38	>
39	>	class DBSInfoError:
40	>	def __init__(self, url):
41	>	print '\nERROR accessing DBS url : '+url+'\n'
42	>	pass
43	>
44
45
6	–	# ####################################
46		class DataDiscoveryError(exceptions.Exception):
47	<	def __init__(self, errorMessage):
48	<	args=errorMessage
49	<	exceptions.Exception.__init__(self, args)
50	<	pass
51	<
52	<	def getErrorMessage(self):
53	<	""" Return exception error """
54	<	return "%s" % (self.args)
47	>	def __init__(self, errorMessage):
48	>	self.args=errorMessage
49	>	exceptions.Exception.__init__(self, self.args)
50	>	pass
51	>
52	>	def getErrorMessage(self):
53	>	""" Return exception error """
54	>	return "%s" % (self.args)
55	>
56	>
57
17	–	# ####################################
58		class NotExistingDatasetError(exceptions.Exception):
59	<	def __init__(self, errorMessage):
60	<	args=errorMessage
61	<	exceptions.Exception.__init__(self, args)
62	<	pass
63	<
64	<	def getErrorMessage(self):
65	<	""" Return exception error """
66	<	return "%s" % (self.args)
59	>	def __init__(self, errorMessage):
60	>	self.args=errorMessage
61	>	exceptions.Exception.__init__(self, self.args)
62	>	pass
63	>
64	>	def getErrorMessage(self):
65	>	""" Return exception error """
66	>	return "%s" % (self.args)
67	>
68	>
69
28	–	# ####################################
70		class NoDataTierinProvenanceError(exceptions.Exception):
71	<	def __init__(self, errorMessage):
72	<	args=errorMessage
73	<	exceptions.Exception.__init__(self, args)
74	<	pass
75	<
76	<	def getErrorMessage(self):
77	<	""" Return exception error """
78	<	return "%s" % (self.args)
71	>	def __init__(self, errorMessage):
72	>	self.args=errorMessage
73	>	exceptions.Exception.__init__(self, self.args)
74	>	pass
75	>
76	>	def getErrorMessage(self):
77	>	""" Return exception error """
78	>	return "%s" % (self.args)
79
39	–	# ####################################
40	–	# class to find and extact info from published data
41	–	class DataDiscovery:
42	–	def __init__(self, owner, dataset, dataTiers, cfg_params):
80
81	<	#       Attributes
82	<	self.owner = owner
83	<	self.dataset = dataset
84	<	self.dataTiers = dataTiers
81	>
82	>	class DataDiscovery:
83	>	"""
84	>	Class to find and extact info from published data
85	>	"""
86	>	def __init__(self, datasetPath, cfg_params, skipAnBlocks):
87	>
88	>	#       Attributes
89	>	self.datasetPath = datasetPath
90	>	# Analysis dataset is primary/processed/tier/definition
91	>	self.ads = len(self.datasetPath.split("/")) > 4
92		self.cfg_params = cfg_params
93	+	self.skipBlocks = skipAnBlocks
94	+
95	+	self.eventsPerBlock = {}  # DBS output: map fileblocks-events for collection
96	+	self.eventsPerFile = {}   # DBS output: map files-events
97	+	#         self.lumisPerBlock = {}   # DBS output: number of lumis in each block
98	+	#         self.lumisPerFile = {}    # DBS output: number of lumis in each file
99	+	self.blocksinfo = {}      # DBS output: map fileblocks-files
100	+	self.maxEvents = 0        # DBS output: max events
101	+	self.maxLumis = 0         # DBS output: total number of lumis
102	+	self.parent = {}          # DBS output: parents of each file
103	+	self.lumis = {}           # DBS output: lumis in each file
104
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
105
55	–	# ####################################
106		def fetchDBSInfo(self):
107		"""
108		Contact DBS
109		"""
110	<
111	<	## add the PU among the required data tiers if the Digi are requested
112	<	if (self.dataTiers.count('Digi')>0) & (self.dataTiers.count('PU')<=0) :
113	<	self.dataTiers.append('PU')
114	<
115	<	## get info about the requested dataset
116	<	dbs=DBSInfo()
110	>	## get DBS URL
111	>	global_url="http://cmsdbsprod.cern.ch/cms_dbs_prod_global/servlet/DBSServlet"
112	>	caf_url = "http://cmsdbsprod.cern.ch/cms_dbs_caf_analysis_01/servlet/DBSServlet"
113	>	dbs_url_map  =   {'glite':    global_url,
114	>	'glitecoll':global_url,\
115	>	'condor':   global_url,\
116	>	'condor_g': global_url,\
117	>	'glidein':  global_url,\
118	>	'lsf':      global_url,\
119	>	'caf':      caf_url,\
120	>	'sge':      global_url,\
121	>	'arc':      global_url,\
122	>	'pbs':      global_url
123	>	}
124	>
125	>	dbs_url_default = dbs_url_map[(common.scheduler.name()).lower()]
126	>	dbs_url=  self.cfg_params.get('CMSSW.dbs_url', dbs_url_default)
127	>	common.logger.info("Accessing DBS at: "+dbs_url)
128	>
129	>	## check if runs are selected
130	>	runselection = []
131	>	if (self.cfg_params.has_key('CMSSW.runselection')):
132	>	runselection = parseRange2(self.cfg_params['CMSSW.runselection'])
133	>
134	>
135	>	self.splitByRun = int(self.cfg_params.get('CMSSW.split_by_run', 0))
136	>
137	>	common.logger.log(10-1,"runselection is: %s"%runselection)
138	>	## service API
139	>	args = {}
140	>	args['url']     = dbs_url
141	>	args['level']   = 'CRITICAL'
142	>
143	>	## check if has been requested to use the parent info
144	>	useparent = int(self.cfg_params.get('CMSSW.use_parent',0))
145	>
146	>	## check if has been asked for a non default file to store/read analyzed fileBlocks
147	>	defaultName = common.work_space.shareDir()+'AnalyzedBlocks.txt'
148	>	fileBlocks_FileName = os.path.abspath(self.cfg_params.get('CMSSW.fileblocks_file',defaultName))
149	>
150	>	api = DBSAPI.dbsApi.DbsApi(args)
151	>	self.files = self.queryDbs(api,path=self.datasetPath,runselection=runselection,useParent=useparent)
152	>
153	>	anFileBlocks = []
154	>	if self.skipBlocks: anFileBlocks = readTXTfile(self, fileBlocks_FileName)
155	>
156	>	# parse files and fill arrays
157	>	for file in self.files :
158	>	parList  = []
159	>	lumiList = [] # List of tuples
160	>	# skip already analyzed blocks
161	>	fileblock = file['Block']['Name']
162	>	if fileblock not in anFileBlocks :
163	>	filename = file['LogicalFileName']
164	>	# asked retry the list of parent for the given child
165	>	if useparent==1:
166	>	parList = [x['LogicalFileName'] for x in file['ParentList']]
167	>	if self.ads:
168	>	lumiList = [ (x['RunNumber'], x['LumiSectionNumber'])
169	>	for x in file['LumiList'] ]
170	>	self.parent[filename] = parList
171	>	self.lumis[filename] = lumiList
172	>	if filename.find('.dat') < 0 :
173	>	events    = file['NumberOfEvents']
174	>	# Count number of events and lumis per block
175	>	if fileblock in self.eventsPerBlock.keys() :
176	>	self.eventsPerBlock[fileblock] += events
177	>	else :
178	>	self.eventsPerBlock[fileblock] = events
179	>	# Number of events per file
180	>	self.eventsPerFile[filename] = events
181	>
182	>	# List of files per block
183	>	if fileblock in self.blocksinfo.keys() :
184	>	self.blocksinfo[fileblock].append(filename)
185	>	else :
186	>	self.blocksinfo[fileblock] = [filename]
187	>
188	>	# total number of events
189	>	self.maxEvents += events
190	>	self.maxLumis  += len(lumiList)
191	>
192	>	if  self.skipBlocks and len(self.eventsPerBlock.keys()) == 0:
193	>	msg = "No new fileblocks available for dataset: "+str(self.datasetPath)
194	>	raise  CrabException(msg)
195	>
196	>	saveFblocks=''
197	>	for block in self.eventsPerBlock.keys() :
198	>	saveFblocks += str(block)+'\n'
199	>	common.logger.log(10-1,"DBSInfo: total nevts %i in block %s "%(self.eventsPerBlock[block],block))
200	>	writeTXTfile(self, fileBlocks_FileName , saveFblocks)
201	>
202	>	if len(self.eventsPerBlock) <= 0:
203	>	raise NotExistingDatasetError(("\nNo data for %s in DBS\nPlease check"
204	>	+ " dataset path variables in crab.cfg")
205	>	% self.datasetPath)
206	>
207	>
208	>	def queryDbs(self,api,path=None,runselection=None,useParent=None):
209	>
210	>	allowedRetriveValue = ['retrive_block', 'retrive_run']
211	>	if self.ads: allowedRetriveValue.append('retrive_lumi')
212	>	if useParent == 1: allowedRetriveValue.append('retrive_parent')
213	>	common.logger.debug("Set of input parameters used for DBS query: %s" % allowedRetriveValue)
214		try:
215	<	self.datasets = dbs.getMatchingDatasets(self.owner, self.dataset)
216	<	except DBSError, ex:
217	<	raise DataDiscoveryError(ex.getErrorMessage())
218	<	if len(self.datasets) == 0:
219	<	raise DataDiscoveryError("Owner=%s, Dataset=%s unknown to DBS" % (self.owner, self.dataset))
220	<	if len(self.datasets) > 1:
221	<	raise DataDiscoveryError("Owner=%s, Dataset=%s is ambiguous" % (self.owner, self.dataset))
222	<	try:
223	<	self.dbsdataset = self.datasets[0].get('datasetPathName')
224	<	self.blocksinfo = dbs.getDatasetContents(self.dbsdataset)
225	<	self.allblocks.append (self.blocksinfo.keys ()) # add also the current fileblocksinfo
226	<	self.dbspaths.append(self.dbsdataset)
227	<	except DBSError, ex:
228	<	raise DataDiscoveryError(ex.getErrorMessage())
229	<
230	<	if len(self.blocksinfo)<=0:
231	<	msg="\nERROR Data for %s do not exist in DBS! \n Check the dataset/owner variables in crab.cfg !"%self.dbsdataset
232	<	raise NotExistingDatasetError(msg)
215	>	if len(runselection) <=0 :
216	>	if useParent==1 or self.splitByRun==1 :
217	>	if self.ads:
218	>	files = api.listFiles(analysisDataset=path, retriveList=allowedRetriveValue)
219	>	else :
220	>	files = api.listFiles(path=path, retriveList=allowedRetriveValue)
221	>	else:
222	>	files = api.listDatasetFiles(self.datasetPath)
223	>	else :
224	>	files=[]
225	>	for arun in runselection:
226	>	try:
227	>	if self.ads:
228	>	filesinrun = api.listFiles(analysisDataset=path,retriveList=allowedRetriveValue,runNumber=arun)
229	>	else:
230	>	filesinrun = api.listFiles(path=path,retriveList=allowedRetriveValue,runNumber=arun)
231	>	files.extend(filesinrun)
232	>	except:
233	>	msg="WARNING: problem extracting info from DBS for run %s "%arun
234	>	common.logger.info(msg)
235	>	pass
236	>
237	>	except DbsBadRequest, msg:
238	>	raise DataDiscoveryError(msg)
239	>	except DBSError, msg:
240	>	raise DataDiscoveryError(msg)
241
242	+	return files
243
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())
244
245	<	## check that the user asks for parent Data Tier really existing in the DBS provenance
246	<	self.checkParentDataTier(parents, self.dataTiers)
245	>	def getMaxEvents(self):
246	>	"""
247	>	max events
248	>	"""
249	>	return self.maxEvents
250
251	<	## for each parent get the corresponding fileblocks
252	<	try:
253	<	for p in parents:
254	<	## fill a list of dbspaths
255	<	parentPath = p.get('parent').get('datasetPathName')
256	<	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)
251	>
252	>	def getMaxLumis(self):
253	>	"""
254	>	Return the number of lumis in the dataset
255	>	"""
256	>	return self.maxLumis
257
258
259	<	# #################################################
126	<	def getMaxEvents(self):
259	>	def getEventsPerBlock(self):
260		"""
261	<	max events of the primary dataset-owner
261	>	list the event collections structure by fileblock
262		"""
263	<	## loop over the fileblocks of the primary dataset-owner
131	<	nevts=0
132	<	for blockevts in self.blocksinfo.values():
133	<	nevts=nevts+blockevts
263	>	return self.eventsPerBlock
264
135	–	return nevts
265
266	<	# #################################################
138	<	def getDBSPaths(self):
266	>	def getEventsPerFile(self):
267		"""
268	<	list the DBSpaths for all required data
268	>	list the event collections structure by file
269		"""
270	<	return self.dbspaths
270	>	return self.eventsPerFile
271
272	<	# #################################################
273	<	def getEVC(self):
272	>
273	>	def getFiles(self):
274		"""
275	<	list the event collections structure by fileblock
275	>	return files grouped by fileblock
276		"""
277	<	print "To be used by a more complex job splitting... TODO later... "
278	<	print "it requires changes in what's returned by DBSInfo.getDatasetContents and then fetchDBSInfo"
277	>	return self.blocksinfo
278	>
279
280	<	# #################################################
153	<	def getFileBlocks(self):
280	>	def getParent(self):
281		"""
282	<	fileblocks for all required dataset-owners
282	>	return parent grouped by file
283		"""
284	<	return self.allblocks
284	>	return self.parent
285
286	<	########################################################################
286	>
287	>	def getLumis(self):
288	>	"""
289	>	return lumi sections grouped by file
290	>	"""
291	>	return self.lumis
292
293
294	+	def getListFiles(self):
295	+	"""
296	+	return parent grouped by file
297	+	"""
298	+	return self.files

Diff Legend

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