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 1.3 by afanfani, Sun Jan 29 01:46:08 2006 UTC vs.
Revision 1.30 by spiga, Tue Mar 31 22:36:42 2009 UTC

#	Line 1 | Line 1
1	<	#!/usr/bin/env python2
2	<	import sys, os, string, re
3	<	from DBSInfo import *
4	<
1	>	#!/usr/bin/env python
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	>	class DBSInvalidDataTierError(exceptions.Exception):
23	>	def __init__(self, errorName, errorMessage):
24	>	args='\nERROR DBS %s : %s \n'%(errorName,errorMessage)
25	>	exceptions.Exception.__init__(self, args)
26	>	pass
27	>
28	>	def getErrorMessage(self):
29	>	""" Return error message """
30	>	return "%s" % (self.args)
31	>
32	>	# #######################################
33	>	class DBSInfoError:
34	>	def __init__(self, url):
35	>	print '\nERROR accessing DBS url : '+url+'\n'
36	>	pass
37
38		# ####################################
39		class DataDiscoveryError(exceptions.Exception):
40	<	def __init__(self, errorMessage):
41	<	args=errorMessage
42	<	exceptions.Exception.__init__(self, args)
43	<	pass
44	<
45	<	def getErrorMessage(self):
46	<	""" Return exception error """
47	<	return "%s" % (self.args)
40	>	def __init__(self, errorMessage):
41	>	self.args=errorMessage
42	>	exceptions.Exception.__init__(self, self.args)
43	>	pass
44	>
45	>	def getErrorMessage(self):
46	>	""" Return exception error """
47	>	return "%s" % (self.args)
48
49		# ####################################
50		class NotExistingDatasetError(exceptions.Exception):
51	<	def __init__(self, errorMessage):
52	<	args=errorMessage
53	<	exceptions.Exception.__init__(self, args)
54	<	pass
55	<
56	<	def getErrorMessage(self):
57	<	""" Return exception error """
58	<	return "%s" % (self.args)
51	>	def __init__(self, errorMessage):
52	>	self.args=errorMessage
53	>	exceptions.Exception.__init__(self, self.args)
54	>	pass
55	>
56	>	def getErrorMessage(self):
57	>	""" Return exception error """
58	>	return "%s" % (self.args)
59
60		# ####################################
61		class NoDataTierinProvenanceError(exceptions.Exception):
62	<	def __init__(self, errorMessage):
63	<	args=errorMessage
64	<	exceptions.Exception.__init__(self, args)
65	<	pass
66	<
67	<	def getErrorMessage(self):
68	<	""" Return exception error """
69	<	return "%s" % (self.args)
62	>	def __init__(self, errorMessage):
63	>	self.args=errorMessage
64	>	exceptions.Exception.__init__(self, self.args)
65	>	pass
66	>
67	>	def getErrorMessage(self):
68	>	""" Return exception error """
69	>	return "%s" % (self.args)
70
71		# ####################################
72		# class to find and extact info from published data
73		class DataDiscovery:
74	<	def __init__(self, owner, dataset, dataTiers, cfg_params):
74	>	def __init__(self, datasetPath, cfg_params, skipAnBlocks):
75
76	<	#       Attributes
77	<	self.dbsdataset='/'+dataset+'/datatier/'+owner
46	<	self.dataTiers = dataTiers
76	>	#       Attributes
77	>	self.datasetPath = datasetPath
78		self.cfg_params = cfg_params
79	+	self.skipBlocks = skipAnBlocks
80
81	<	self.dbspaths= []     # DBS output: list of dbspaths for all data
82	<	self.allblocks = []   # DBS output: list of map fileblocks-totevts for all dataset-owners
83	<	self.blocksinfo = {}  # DBS output: map fileblocks-totevts for the primary block, used internally to this class
84	<	#DBS output: max events computed by method getMaxEvents
81	>	self.eventsPerBlock = {}  # DBS output: map fileblocks-events for collection
82	>	self.eventsPerFile = {}   # DBS output: map files-events
83	>	self.blocksinfo = {}      # DBS output: map fileblocks-files
84	>	self.maxEvents = 0        # DBS output: max events
85	>	self.parent = {}       # DBS output: max events
86
87		# ####################################
88		def fetchDBSInfo(self):
89		"""
90		Contact DBS
91		"""
92	<	parents = []
93	<	parentsblocksinfo = {}
94	<
95	<	## add the PU among the required data tiers if the Digi are requested
96	<	if (self.dataTiers.count('Digi')>0) & (self.dataTiers.count('PU')<=0) :
97	<	self.dataTiers.append('PU')
98	<
99	<	## get info about the requested dataset
100	<	dbs=DBSInfo(self.dbsdataset,self.dataTiers)
101	<	try:
102	<	self.blocksinfo=dbs.getDatasetContents()
103	<	except DBSError, ex:
104	<	raise DataDiscoveryError(ex.getErrorMessage())
105	<
106	<	if len(self.blocksinfo)<=0:
107	<	msg="\nERROR Data %s do not exist in DBS! \n Check the dataset/owner variables in crab.cfg !"%self.dbsdataset
108	<	raise NotExistingDatasetError(msg)
109	<
110	<	currentdatatier=string.split(self.blocksinfo.keys()[0],'/')[2]
111	<	fakedatatier=string.split(self.dbsdataset,'/')[2]
112	<	currentdbsdataset=string.replace(self.dbsdataset, fakedatatier, currentdatatier)
113	<
114	<	self.dbspaths.append(currentdbsdataset)    # add the requested dbspath
115	<
116	<	## get info about the parents
92	>	## get DBS URL
93	>	global_url="http://cmsdbsprod.cern.ch/cms_dbs_prod_global/servlet/DBSServlet"
94	>	caf_url = "http://cmsdbsprod.cern.ch/cms_dbs_caf_analysis_01/servlet/DBSServlet"
95	>	dbs_url_map  =   {'glite':    global_url,
96	>	'glitecoll':global_url,\
97	>	'condor':   global_url,\
98	>	'condor_g': global_url,\
99	>	'glidein':  global_url,\
100	>	'lsf':      global_url,\
101	>	'caf':      caf_url,\
102	>	'sge':      global_url,
103	>	'arc':      global_url
104	>	}
105	>
106	>	dbs_url_default = dbs_url_map[(common.scheduler.name()).lower()]
107	>	dbs_url=  self.cfg_params.get('CMSSW.dbs_url', dbs_url_default)
108	>	common.logger.debug(3,"Accessing DBS at: "+dbs_url)
109	>
110	>	## check if runs are selected
111	>	runselection = []
112	>	if (self.cfg_params.has_key('CMSSW.runselection')):
113	>	runselection = parseRange2(self.cfg_params['CMSSW.runselection'])
114	>
115	>
116	>	self.splitByRun = int(self.cfg_params.get('CMSSW.split_by_run', 0))
117	>
118	>	self.ads = int(self.cfg_params.get('CMSSW.ads', 0))
119	>
120	>	common.logger.debug(6,"runselection is: %s"%runselection)
121	>	## service API
122	>	args = {}
123	>	args['url']     = dbs_url
124	>	args['level']   = 'CRITICAL'
125	>
126	>	## check if has been requested to use the parent info
127	>	useparent = int(self.cfg_params.get('CMSSW.use_parent',0))
128	>
129	>	## check if has been asked for a non default file to store/read analyzed fileBlocks
130	>	defaultName = common.work_space.shareDir()+'AnalyzedBlocks.txt'
131	>	fileBlocks_FileName = os.path.abspath(self.cfg_params.get('CMSSW.fileblocks_file',defaultName))
132	>
133	>	api = DBSAPI.dbsApi.DbsApi(args)
134	>
135	>	self.files = self.queryDbs(api,path=self.datasetPath,runselection=runselection,useParent=useparent)
136	>
137	>	anFileBlocks = []
138	>	if self.skipBlocks: anFileBlocks = readTXTfile(self, fileBlocks_FileName)
139	>
140	>	# parse files and fill arrays
141	>	for file in self.files :
142	>	parList = []
143	>	# skip already analyzed blocks
144	>	fileblock = file['Block']['Name']
145	>	if fileblock not in anFileBlocks :
146	>	filename = file['LogicalFileName']
147	>	# asked retry the list of parent for the given child
148	>	if useparent==1: parList = [x['LogicalFileName'] for x in file['ParentList']]
149	>	self.parent[filename] = parList
150	>	if filename.find('.dat') < 0 :
151	>	events    = file['NumberOfEvents']
152	>	# number of events per block
153	>	if fileblock in self.eventsPerBlock.keys() :
154	>	self.eventsPerBlock[fileblock] += events
155	>	else :
156	>	self.eventsPerBlock[fileblock] = events
157	>	# number of events per file
158	>	self.eventsPerFile[filename] = events
159	>
160	>	# number of events per block
161	>	if fileblock in self.blocksinfo.keys() :
162	>	self.blocksinfo[fileblock].append(filename)
163	>	else :
164	>	self.blocksinfo[fileblock] = [filename]
165	>
166	>	# total number of events
167	>	self.maxEvents += events
168	>	if  self.skipBlocks and len(self.eventsPerBlock.keys()) == 0:
169	>	msg = "No new fileblocks available for dataset: "+str(self.datasetPath)
170	>	raise  CrabException(msg)
171	>
172	>	saveFblocks=''
173	>	for block in self.eventsPerBlock.keys() :
174	>	saveFblocks += str(block)+'\n'
175	>	common.logger.debug(6,"DBSInfo: total nevts %i in block %s "%(self.eventsPerBlock[block],block))
176	>	writeTXTfile(self, fileBlocks_FileName , saveFblocks)
177	>
178	>	if len(self.eventsPerBlock) <= 0:
179	>	raise NotExistingDatasetError(("\nNo data for %s in DBS\nPlease check"
180	>	+ " dataset path variables in crab.cfg")
181	>	% self.datasetPath)
182	>
183	>
184	>	###########################
185	>
186	>	def queryDbs(self,api,path=None,runselection=None,useParent=None):
187	>
188	>	allowedRetriveValue = ['retrive_block', 'retrive_run']
189	>	if useParent==1 : allowedRetriveValue = allowedRetriveValue + ['retrive_parent']
190		try:
191	<	parents=dbs.getDatasetProvenance()
192	<	except DBSInvalidDataTierError, ex:
193	<	msg=ex.getErrorMessage()+' \n Check the data_tier variable in crab.cfg !\n'
194	<	raise DataDiscoveryError(msg)
195	<	except DBSError, ex:
196	<	raise DataDiscoveryError(ex.getErrorMessage())
197	<
198	<	## check that the user asks for parent Data Tier really existing in the DBS provenance
199	<	self.checkParentDataTier(parents, self.dataTiers, currentdbsdataset)
200	<
201	<	## for each parent get the corresponding fileblocks
202	<	for aparent in parents:
203	<	## fill a list of dbspaths
204	<	parentdbsdataset=aparent.getDatasetPath()
205	<	self.dbspaths.append(parentdbsdataset)
206	<	pdbs=DBSInfo(parentdbsdataset,[])
207	<	try:
208	<	parentsblocksinfo=pdbs.getDatasetContents()
209	<	except DBSError, ex:
210	<	raise DataDiscoveryError(ex.getErrorMessage())
211	<
212	<	self.allblocks.append(parentsblocksinfo.keys()) # add parent fileblocksinfo
213	<
214	<	## all the required blocks
215	<	self.allblocks.append(self.blocksinfo.keys()) # add also the current fileblocksinfo
191	>	if len(runselection) <=0 :
192	>	if useParent==1 or self.splitByRun==1 :
193	>	if self.ads==1 :
194	>	files = api.listFiles(analysisDataset=path, retriveList=allowedRetriveValue)
195	>	else :
196	>	files = api.listFiles(path=path, retriveList=allowedRetriveValue)
197	>	common.logger.debug(5,"Set of input parameters used for DBS query : \n"+str(allowedRetriveValue))
198	>	common.logger.write("Set of input parameters used for DBS query : \n"+str(allowedRetriveValue))
199	>	else:
200	>	files = api.listDatasetFiles(self.datasetPath)
201	>	else :
202	>	files=[]
203	>	for arun in runselection:
204	>	try:
205	>	if self.ads==1 : filesinrun = api.listFiles(analysisDataset=path,retriveList=allowedRetriveValue,runNumber=arun)
206	>	else: filesinrun = api.listFiles(path=path,retriveList=allowedRetriveValue,runNumber=arun)
207	>	files.extend(filesinrun)
208	>	except:
209	>	msg="WARNING: problem extracting info from DBS for run %s "%arun
210	>	common.logger.message(msg)
211	>	pass
212	>
213	>	except DbsBadRequest, msg:
214	>	raise DataDiscoveryError(msg)
215	>	except DBSError, msg:
216	>	raise DataDiscoveryError(msg)
217
218	+	return files
219
220		# #################################################
221	<	def checkParentDataTier(self, parents, user_datatiers, currentdbsdataset ):
221	>	def getMaxEvents(self):
222		"""
223	<	check that the data tiers requested by the user really exists in the provenance of the given dataset
223	>	max events
224		"""
225	<
118	<	current_datatier=string.split(currentdbsdataset,'/')[2]
119	<
120	<	parent_datatypes=[]
121	<	for aparent in parents:
122	<	parent_datatypes.append(aparent.getDataType())
123	<
124	<	for datatier in user_datatiers:
125	<	if parent_datatypes.count(datatier)<=0:
126	<	# the current datatier is not supposed to be in the provenance
127	<	if not (datatier == current_datatier):
128	<	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 !"%(currentdbsdataset,datatier)
129	<	raise  NoDataTierinProvenanceError(msg)
130	<
225	>	return self.maxEvents
226
227		# #################################################
228	<	def getMaxEvents(self):
228	>	def getEventsPerBlock(self):
229		"""
230	<	max events of the primary dataset-owner
230	>	list the event collections structure by fileblock
231		"""
232	<	## loop over the fileblocks of the primary dataset-owner
138	<	nevts=0
139	<	for blockevts in self.blocksinfo.values():
140	<	nevts=nevts+blockevts
232	>	return self.eventsPerBlock
233
234	<	return nevts
234	>	# #################################################
235	>	def getEventsPerFile(self):
236	>	"""
237	>	list the event collections structure by file
238	>	"""
239	>	return self.eventsPerFile
240
241		# #################################################
242	<	def getDBSPaths(self):
242	>	def getFiles(self):
243		"""
244	<	list the DBSpaths for all required data
244	>	return files grouped by fileblock
245		"""
246	<	return self.dbspaths
246	>	return self.blocksinfo
247
248		# #################################################
249	<	def getEVC(self):
249	>	def getParent(self):
250		"""
251	<	list the event collections structure by fileblock
251	>	return parent grouped by file
252		"""
253	<	print "To be used by a more complex job splitting... TODO later... "
157	<	print "it requires changes in what's returned by DBSInfo.getDatasetContents and then fetchDBSInfo"
253	>	return self.parent
254
255		# #################################################
256	<	def getFileBlocks(self):
256	>	def getListFiles(self):
257		"""
258	<	fileblocks for all required dataset-owners
258	>	return parent grouped by file
259		"""
260	<	return self.allblocks
260	>	return self.files
261
262		########################################################################
167	–
168	–

Diff Legend

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