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 1.33 by ewv, Thu Jul 30 18:45:44 2009 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	>	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)
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	<	# ####################################
56	>
57	>
58		class NotExistingDatasetError(exceptions.Exception):
59		def __init__(self, errorMessage):
60	<	args=errorMessage
61	<	exceptions.Exception.__init__(self, args)
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	<	# ####################################
68	>
69	>
70		class NoDataTierinProvenanceError(exceptions.Exception):
71		def __init__(self, errorMessage):
72	<	args=errorMessage
73	<	exceptions.Exception.__init__(self, args)
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("/")) > 3
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	+	## 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	+	}
123	+
124	+	dbs_url_default = dbs_url_map[(common.scheduler.name()).lower()]
125	+	dbs_url=  self.cfg_params.get('CMSSW.dbs_url', dbs_url_default)
126	+	common.logger.debug("Accessing DBS at: "+dbs_url)
127	+
128	+	## check if runs are selected
129	+	runselection = []
130	+	if (self.cfg_params.has_key('CMSSW.runselection')):
131	+	runselection = parseRange2(self.cfg_params['CMSSW.runselection'])
132	+
133	+
134	+	self.splitByRun = int(self.cfg_params.get('CMSSW.split_by_run', 0))
135	+
136	+	common.logger.log(10-1,"runselection is: %s"%runselection)
137	+	## service API
138	+	args = {}
139	+	args['url']     = dbs_url
140	+	args['level']   = 'CRITICAL'
141	+
142	+	## check if has been requested to use the parent info
143	+	useparent = int(self.cfg_params.get('CMSSW.use_parent',0))
144	+
145	+	## check if has been asked for a non default file to store/read analyzed fileBlocks
146	+	defaultName = common.work_space.shareDir()+'AnalyzedBlocks.txt'
147	+	fileBlocks_FileName = os.path.abspath(self.cfg_params.get('CMSSW.fileblocks_file',defaultName))
148	+
149	+	api = DBSAPI.dbsApi.DbsApi(args)
150	+	self.files = self.queryDbs(api,path=self.datasetPath,runselection=runselection,useParent=useparent)
151	+
152	+	anFileBlocks = []
153	+	if self.skipBlocks: anFileBlocks = readTXTfile(self, fileBlocks_FileName)
154	+
155	+	# parse files and fill arrays
156	+	for file in self.files :
157	+	parList  = []
158	+	lumiList = [] # List of tuples
159	+	# skip already analyzed blocks
160	+	fileblock = file['Block']['Name']
161	+	if fileblock not in anFileBlocks :
162	+	filename = file['LogicalFileName']
163	+	# asked retry the list of parent for the given child
164	+	if useparent==1:
165	+	parList = [x['LogicalFileName'] for x in file['ParentList']]
166	+	if self.ads:
167	+	lumiList = [ (x['RunNumber'], x['LumiSectionNumber'])
168	+	for x in file['LumiList'] ]
169	+	self.parent[filename] = parList
170	+	self.lumis[filename] = lumiList
171	+	if filename.find('.dat') < 0 :
172	+	events    = file['NumberOfEvents']
173	+	# Count number of events and lumis per block
174	+	if fileblock in self.eventsPerBlock.keys() :
175	+	self.eventsPerBlock[fileblock] += events
176	+	else :
177	+	self.eventsPerBlock[fileblock] = events
178	+	# Number of events per file
179	+	self.eventsPerFile[filename] = events
180	+
181	+	# List of files per block
182	+	if fileblock in self.blocksinfo.keys() :
183	+	self.blocksinfo[fileblock].append(filename)
184	+	else :
185	+	self.blocksinfo[fileblock] = [filename]
186	+
187	+	# total number of events
188	+	self.maxEvents += events
189	+	self.maxLumis  += len(lumiList)
190	+
191	+	if  self.skipBlocks and len(self.eventsPerBlock.keys()) == 0:
192	+	msg = "No new fileblocks available for dataset: "+str(self.datasetPath)
193	+	raise  CrabException(msg)
194	+
195	+	saveFblocks=''
196	+	for block in self.eventsPerBlock.keys() :
197	+	saveFblocks += str(block)+'\n'
198	+	common.logger.log(10-1,"DBSInfo: total nevts %i in block %s "%(self.eventsPerBlock[block],block))
199	+	writeTXTfile(self, fileBlocks_FileName , saveFblocks)
200	+
201	+	if len(self.eventsPerBlock) <= 0:
202	+	raise NotExistingDatasetError(("\nNo data for %s in DBS\nPlease check"
203	+	+ " dataset path variables in crab.cfg")
204	+	% self.datasetPath)
205	+
206	+
207	+	def queryDbs(self,api,path=None,runselection=None,useParent=None):
208	+
209	+	allowedRetriveValue = ['retrive_block', 'retrive_run']
210	+	if self.ads: allowedRetriveValue.append('retrive_lumi')
211	+	if useParent == 1: allowedRetriveValue.append('retrive_parent')
212	+	common.logger.debug("Set of input parameters used for DBS query: %s" % allowedRetriveValue)
213	+	try:
214	+	if len(runselection) <=0 :
215	+	if useParent==1 or self.splitByRun==1 :
216	+	if self.ads:
217	+	files = api.listFiles(analysisDataset=path, retriveList=allowedRetriveValue)
218	+	else :
219	+	files = api.listFiles(path=path, retriveList=allowedRetriveValue)
220	+	else:
221	+	files = api.listDatasetFiles(self.datasetPath)
222	+	else :
223	+	files=[]
224	+	for arun in runselection:
225	+	try:
226	+	if self.ads:
227	+	filesinrun = api.listFiles(analysisDataset=path,retriveList=allowedRetriveValue,runNumber=arun)
228	+	else:
229	+	filesinrun = api.listFiles(path=path,retriveList=allowedRetriveValue,runNumber=arun)
230	+	files.extend(filesinrun)
231	+	except:
232	+	msg="WARNING: problem extracting info from DBS for run %s "%arun
233	+	common.logger.info(msg)
234	+	pass
235
236	<	## add the PU among the required data tiers if the Digi are requested
237	<	if (self.dataTiers.count('Digi')>0) & (self.dataTiers.count('PU')<=0) :
238	<	self.dataTiers.append('PU')
236	>	except DbsBadRequest, msg:
237	>	raise DataDiscoveryError(msg)
238	>	except DBSError, msg:
239	>	raise DataDiscoveryError(msg)
240
241	<	## 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)
241	>	return files
242
243
244	<	## get info about the parents
245	<	try:
246	<	parents=dbs.getDatasetProvenance(self.dbsdataset, self.dataTiers)
247	<	except DBSInvalidDataTierError, ex:
248	<	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	>	def getMaxEvents(self):
245	>	"""
246	>	max events
247	>	"""
248	>	return self.maxEvents
249
97	–	## check that the user asks for parent Data Tier really existing in the DBS provenance
98	–	self.checkParentDataTier(parents, self.dataTiers)
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')
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)
251	>	def getMaxLumis(self):
252	>	"""
253	>	Return the number of lumis in the dataset
254	>	"""
255	>	return self.maxLumis
256
257
258	<	# #################################################
259	<	def getMaxEvents(self):
258	>	def getEventsPerBlock(self):
259	>	"""
260	>	list the event collections structure by fileblock
261	>	"""
262	>	return self.eventsPerBlock
263	>
264	>
265	>	def getEventsPerFile(self):
266		"""
267	<	max events of the primary dataset-owner
267	>	list the event collections structure by file
268		"""
269	<	## loop over the fileblocks of the primary dataset-owner
131	<	nevts=0
132	<	for blockevts in self.blocksinfo.values():
133	<	nevts=nevts+blockevts
269	>	return self.eventsPerFile
270
135	–	return nevts
271
272	<	# #################################################
138	<	def getDBSPaths(self):
272	>	def getFiles(self):
273		"""
274	<	list the DBSpaths for all required data
274	>	return files grouped by fileblock
275		"""
276	<	return self.dbspaths
276	>	return self.blocksinfo
277
278	<	# #################################################
279	<	def getEVC(self):
278	>
279	>	def getParent(self):
280		"""
281	<	list the event collections structure by fileblock
281	>	return parent grouped by file
282		"""
283	<	print "To be used by a more complex job splitting... TODO later... "
284	<	print "it requires changes in what's returned by DBSInfo.getDatasetContents and then fetchDBSInfo"
283	>	return self.parent
284	>
285
286	<	# #################################################
153	<	def getFileBlocks(self):
286	>	def getLumis(self):
287		"""
288	<	fileblocks for all required dataset-owners
288	>	return lumi sections grouped by file
289		"""
290	<	return self.allblocks
290	>	return self.lumis
291
292	<	########################################################################
292	>
293	>	def getListFiles(self):
294	>	"""
295	>	return parent grouped by file
296	>	"""
297	>	return self.files

Diff Legend

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