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 1.1 by afanfani, Mon Nov 7 23:17:44 2005 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	<	# ####################################
4	<	class DataDiscoveryError:
5	<	def __init__(self):
6	<	print '\nERROR accessing Data Discovery\n'
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	<	class DatasetContentsError:
22	<	def __init__(self):
23	<	print '\nERROR accessing Data Discovery : getDatasetContents\n'
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	<	# ####################################
34	<	class DatasetProvenanceError:
35	<	def __init__(self):
36	<	print '\nERROR accessing Data Discovery : getDatasetProvenance\n'
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
22	–	# ####################################
23	–	# class to find and extact info from published data
24	–	class DataDiscovery:
25	–	def __init__(self, owner, dataset, dataTiers, cfg_params):
44
45	<	#       Attributes
46	<	self.dbsdataset=dataset+'/datatier/'+owner
47	<	self.dataTiers = dataTiers
45	>
46	>	class DataDiscoveryError(exceptions.Exception):
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	>
58	>	class NotExistingDatasetError(exceptions.Exception):
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	>
70	>	class NoDataTierinProvenanceError(exceptions.Exception):
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	>
80	>
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
32	–	self.dbspaths= []  # DBS output: list of dbspaths for all data
33	–	self.allblocks = []   # DBS output: list of map fileblocks-totevts for all dataset-owners
34	–	self.blocksinfo = {}     # DBS output: map fileblocks-totevts for the primary block, used internally to this class
35	–	#DBS output: max events computed by method getMaxEvents
105
37	–	# ####################################
106		def fetchDBSInfo(self):
107		"""
108		Contact DBS
109		"""
110	<	parents = []
111	<	parentsblocksinfo = {}
112	<	self.dbspaths.append("/"+self.dbsdataset) # add the primary dbspath
113	<	# it might be replaced if one get from DBSAPI the primary dbspath as well
114	<
115	<	dbs=DBSInfo(self.dbsdataset,self.dataTiers)
116	<	try:
117	<	self.blocksinfo=dbs.getDatasetContents()
118	<	except dbs.DBSError:
119	<	raise DataDiscoveryError
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	<	parents=dbs.getDatasetProvenance()
216	<	except:
217	<	raise DataDiscoveryError
218	<
219	<	## for each parent get the corresponding fileblocks
220	<	for aparent in parents:
221	<	## fill the map dataset-owner for the parents
222	<	#pdataset=string.split(aparent,'/')[1]
223	<	#powner=string.split(aparent,'/')[3]
224	<	#self.dataset_owner[powner]=pdataset
225	<	## instead of the map dataset-owner use the dbspaths
226	<	parentdbsdataset=aparent.getDatasetPath()
227	<	self.dbspaths.append(parentdbsdataset)
228	<	#self.dbspaths.append(aparent)
229	<	## get the fileblocks of the parents : FIXME remove the first / in the path
230	<	pdbs=DBSInfo(parentdbsdataset[1:-1],[])
231	<	try:
232	<	parentsblocksinfo=pdbs.getDatasetContents()
233	<	except:
234	<	raise DataDiscoveryError
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	<	self.allblocks.append(parentsblocksinfo.keys()) # add parent fileblocksinfo
242	>	return files
243
76	–	## all the required blocks
77	–	self.allblocks.append(self.blocksinfo.keys()) # add also the primary fileblocksinfo
244
79	–
80	–	# #################################################
245		def getMaxEvents(self):
246		"""
247	<	max events of the primary dataset-owner
247	>	max events
248		"""
249	<	## loop over the fileblocks of the primary dataset-owner
86	<	nevts=0
87	<	for blockevts in self.blocksinfo.values():
88	<	nevts=nevts+blockevts
249	>	return self.maxEvents
250
90	–	return nevts
251
252	<	# #################################################
93	<	def getDatasetOwnerPairs(self):
252	>	def getMaxLumis(self):
253		"""
254	<	list all required dataset-owner pairs
254	>	Return the number of lumis in the dataset
255		"""
256	<	return self.dataset_owner
257	<	# #################################################
258	<	def getDBSPaths(self):
256	>	return self.maxLumis
257	>
258	>
259	>	def getEventsPerBlock(self):
260		"""
261	<	list the DBSpaths for all required data
261	>	list the event collections structure by fileblock
262		"""
263	<	return self.dbspaths
263	>	return self.eventsPerBlock
264
265	<	# #################################################
266	<	def getEVC(self):
265	>
266	>	def getEventsPerFile(self):
267		"""
268	<	list the event collections structure by fileblock
268	>	list the event collections structure by file
269		"""
270	<	print "To be used by a more complex job splitting... TODO later... "
271	<	print "it requires changes in what's returned by DBSInfo.getDatasetContents and then fetchDBSInfo"
270	>	return self.eventsPerFile
271	>
272
273	<	# #################################################
114	<	def getFileBlocks(self):
273	>	def getFiles(self):
274		"""
275	<	fileblocks for all required dataset-owners
275	>	return files grouped by fileblock
276		"""
277	<	return self.allblocks
277	>	return self.blocksinfo
278	>
279
280	<	########################################################################
280	>	def getParent(self):
281	>	"""
282	>	return parent grouped by file
283	>	"""
284	>	return self.parent
285
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