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 1.14 by slacapra, Wed Jan 17 18:17:58 2007 UTC vs.
Revision 1.37.2.1 by ewv, Thu Jan 21 16:13:28 2010 UTC

#	Line 1 | Line 1
1		#!/usr/bin/env python
2	<	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
5	–	# ####################################
46		class DataDiscoveryError(exceptions.Exception):
47		def __init__(self, errorMessage):
8	–	exceptions.Exception.__init__(self, 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):
19	–	exceptions.Exception.__init__(self, 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):
30	–	exceptions.Exception.__init__(self, 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
80	<	# ####################################
81	<	# class to find and extact info from published data
80	>
81	>
82		class DataDiscovery:
83	<	def __init__(self, datasetPath, dataTiers, cfg_params):
83	>	"""
84	>	Class to find and extact info from published data
85	>	"""
86	>	def __init__(self, datasetPath, cfg_params, skipAnBlocks):
87
88	<	#       Attributes
88	>	#       Attributes
89		self.datasetPath = datasetPath
90	<	self.dataTiers = dataTiers
90	>	# Analysis dataset is primary/processed/tier/definition
91	>	self.ads = len(self.datasetPath.split("/")) > 4 or len(self.datasetPath.split("/")) == 1
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.blocksinfo = {}  # DBS output: map fileblocks-files
98	<	#DBS output: max events computed by method getMaxEvents
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	>
105
53	–	# ####################################
106		def fetchDBSInfo(self):
107		"""
108		Contact DBS
109		"""
58	–
110		## get DBS URL
111	<	try:
112	<	dbs_url=self.cfg_params['CMSSW.dbs_url']
113	<	except KeyError:
114	<	dbs_url="http://cmsdoc.cern.ch/cms/test/aprom/DBS/CGIServer/prodquery"
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	>	'glite_slc5':global_url,\
115	>	'glitecoll':global_url,\
116	>	'condor':   global_url,\
117	>	'condor_g': global_url,\
118	>	'glidein':  global_url,\
119	>	'lsf':      global_url,\
120	>	'caf':      caf_url,\
121	>	'sge':      global_url,\
122	>	'arc':      global_url,\
123	>	'pbs':      global_url
124	>	}
125	>
126	>	dbs_url_default = dbs_url_map[(common.scheduler.name()).lower()]
127	>	dbs_url=  self.cfg_params.get('CMSSW.dbs_url', dbs_url_default)
128	>	common.logger.info("Accessing DBS at: "+dbs_url)
129	>
130	>	## check if runs are selected
131	>	runselection = []
132	>	if (self.cfg_params.has_key('CMSSW.runselection')):
133	>	runselection = parseRange2(self.cfg_params['CMSSW.runselection'])
134	>
135	>
136	>	self.splitByRun = int(self.cfg_params.get('CMSSW.split_by_run', 0))
137	>
138	>	common.logger.log(10-1,"runselection is: %s"%runselection)
139	>	## service API
140	>	args = {}
141	>	args['url']     = dbs_url
142	>	args['level']   = 'CRITICAL'
143	>	args['adshome']   = '$HOME/DBSADS'
144	>
145	>	## check if has been requested to use the parent info
146	>	useparent = int(self.cfg_params.get('CMSSW.use_parent',0))
147	>
148	>	## check if has been asked for a non default file to store/read analyzed fileBlocks
149	>	defaultName = common.work_space.shareDir()+'AnalyzedBlocks.txt'
150	>	fileBlocks_FileName = os.path.abspath(self.cfg_params.get('CMSSW.fileblocks_file',defaultName))
151	>
152	>	api = DBSAPI.dbsApi.DbsApi(args)
153	>	self.files = self.queryDbs(api,path=self.datasetPath,runselection=runselection,useParent=useparent)
154	>
155	>	anFileBlocks = []
156	>	if self.skipBlocks: anFileBlocks = readTXTfile(self, fileBlocks_FileName)
157	>
158	>	# parse files and fill arrays
159	>	for file in self.files :
160	>	parList  = []
161	>	lumiList = [] # List of tuples
162	>	# skip already analyzed blocks
163	>	fileblock = file['Block']['Name']
164	>	if fileblock not in anFileBlocks :
165	>	filename = file['LogicalFileName']
166	>	# asked retry the list of parent for the given child
167	>	if useparent==1:
168	>	parList = [x['LogicalFileName'] for x in file['ParentList']]
169	>	if self.ads:
170	>	lumiList = [ (x['RunNumber'], x['LumiSectionNumber'])
171	>	for x in file['LumiList'] ]
172	>	self.parent[filename] = parList
173	>	self.lumis[filename] = lumiList
174	>	if filename.find('.dat') < 0 :
175	>	events    = file['NumberOfEvents']
176	>	# Count number of events and lumis per block
177	>	if fileblock in self.eventsPerBlock.keys() :
178	>	self.eventsPerBlock[fileblock] += events
179	>	else :
180	>	self.eventsPerBlock[fileblock] = events
181	>	# Number of events per file
182	>	self.eventsPerFile[filename] = events
183	>
184	>	# List of files per block
185	>	if fileblock in self.blocksinfo.keys() :
186	>	self.blocksinfo[fileblock].append(filename)
187	>	else :
188	>	self.blocksinfo[fileblock] = [filename]
189	>
190	>	# total number of events
191	>	self.maxEvents += events
192	>	self.maxLumis  += len(lumiList)
193	>
194	>	if  self.skipBlocks and len(self.eventsPerBlock.keys()) == 0:
195	>	msg = "No new fileblocks available for dataset: "+str(self.datasetPath)
196	>	raise  CrabException(msg)
197	>
198	>	saveFblocks=''
199	>	for block in self.eventsPerBlock.keys() :
200	>	saveFblocks += str(block)+'\n'
201	>	common.logger.log(10-1,"DBSInfo: total nevts %i in block %s "%(self.eventsPerBlock[block],block))
202	>	writeTXTfile(self, fileBlocks_FileName , saveFblocks)
203
204	<	## get info about the requested dataset
205	<	try:
206	<	dbs_instance=self.cfg_params['CMSSW.dbs_instance']
207	<	except KeyError:
208	<	dbs_instance="MCGlobal/Writer"
209	<
210	<	dbs = DBSInfo(dbs_url, dbs_instance)
204	>	if len(self.eventsPerBlock) <= 0:
205	>	raise NotExistingDatasetError(("\nNo data for %s in DBS\nPlease check"
206	>	+ " dataset path variables in crab.cfg")
207	>	% self.datasetPath)
208	>
209	>
210	>	def queryDbs(self,api,path=None,runselection=None,useParent=None):
211	>
212	>	allowedRetriveValue = ['retrive_block', 'retrive_run']
213	>	if self.ads: allowedRetriveValue.append('retrive_lumi')
214	>	if useParent == 1: allowedRetriveValue.append('retrive_parent')
215	>	common.logger.debug("Set of input parameters used for DBS query: %s" % allowedRetriveValue)
216		try:
217	<	self.datasets = dbs.getMatchingDatasets(self.datasetPath)
218	<	except dbsCgiApi.DbsCgiExecutionError, msg:
217	>	if len(runselection) <=0 :
218	>	if useParent==1 or self.splitByRun==1 or self.ads:
219	>	if self.ads:
220	>	files = api.listFiles(analysisDataset=path, retriveList=allowedRetriveValue)
221	>	else :
222	>	files = api.listFiles(path=path, retriveList=allowedRetriveValue)
223	>	else:
224	>	files = api.listDatasetFiles(self.datasetPath)
225	>	else :
226	>	files=[]
227	>	for arun in runselection:
228	>	try:
229	>	if self.ads:
230	>	filesinrun = api.listFiles(analysisDataset=path,retriveList=allowedRetriveValue,runNumber=arun)
231	>	else:
232	>	filesinrun = api.listFiles(path=path,retriveList=allowedRetriveValue,runNumber=arun)
233	>	files.extend(filesinrun)
234	>	except:
235	>	msg="WARNING: problem extracting info from DBS for run %s "%arun
236	>	common.logger.info(msg)
237	>	pass
238	>
239	>	except DbsBadRequest, msg:
240		raise DataDiscoveryError(msg)
241		except DBSError, msg:
242		raise DataDiscoveryError(msg)
243
244	<	if len(self.datasets) == 0:
80	<	raise DataDiscoveryError("DatasetPath=%s unknown to DBS" %self.datasetPath)
81	<	if len(self.datasets) > 1:
82	<	raise DataDiscoveryError("DatasetPath=%s is ambiguous" %self.datasetPath)
244	>	return files
245
84	–	try:
85	–	self.dbsdataset = self.datasets[0].get('datasetPathName')
246
247	<	self.eventsPerBlock = dbs.getEventsPerBlock(self.dbsdataset)
248	<	self.blocksinfo = dbs.getDatasetFileBlocks(self.dbsdataset)
249	<	self.eventsPerFile = dbs.getEventsPerFile(self.dbsdataset)
250	<	except DBSError, ex:
251	<	raise DataDiscoveryError(ex.getErrorMessage())
92	<
93	<	if len(self.eventsPerBlock) <= 0:
94	<	raise NotExistingDatasetError (("\nNo data for %s in DBS\nPlease check"
95	<	+ " dataset path variables in crab.cfg")
96	<	% self.dbsdataset)
247	>	def getMaxEvents(self):
248	>	"""
249	>	max events
250	>	"""
251	>	return self.maxEvents
252
253
254	<	# #################################################
100	<	def getMaxEvents(self):
254	>	def getMaxLumis(self):
255		"""
256	<	max events
256	>	Return the number of lumis in the dataset
257		"""
258	<	## loop over the event collections
105	<	nevts=0
106	<	for evc_evts in self.eventsPerBlock.values():
107	<	nevts=nevts+evc_evts
258	>	return self.maxLumis
259
109	–	return nevts
260
111	–	# #################################################
261		def getEventsPerBlock(self):
262		"""
263	<	list the event collections structure by fileblock
263	>	list the event collections structure by fileblock
264		"""
265		return self.eventsPerBlock
266
267	<	# #################################################
267	>
268		def getEventsPerFile(self):
269		"""
270	<	list the event collections structure by file
270	>	list the event collections structure by file
271		"""
272		return self.eventsPerFile
273
274	<	# #################################################
274	>
275		def getFiles(self):
276		"""
277	<	return files grouped by fileblock
277	>	return files grouped by fileblock
278	>	"""
279	>	return self.blocksinfo
280	>
281	>
282	>	def getParent(self):
283	>	"""
284	>	return parent grouped by file
285		"""
286	<	return self.blocksinfo
286	>	return self.parent
287
288	<	########################################################################
288	>
289	>	def getLumis(self):
290	>	"""
291	>	return lumi sections grouped by file
292	>	"""
293	>	return self.lumis
294	>
295	>
296	>	def getListFiles(self):
297	>	"""
298	>	return parent grouped by file
299	>	"""
300	>	return self.files

Diff Legend

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