CMS 3D CMS Logo
cmsswVersionTools.py
Go to the documentation of this file.
1 import FWCore.ParameterSet.Config as cms
2 
3 from FWCore.GuiBrowsers.ConfigToolBase import *
4 from PhysicsTools.PatAlgos.tools.helpers import *
5 from PhysicsTools.PatAlgos.tools.jetTools import *
6 from Configuration.AlCa.autoCond import autoCond
7 import Utilities.General.cmssw_das_client as das_client
8 import os
9 import socket
10 
11 
12 ## ------------------------------------------------------
13 ## Automatic pick-up of RelVal input files
14 ## ------------------------------------------------------
15 
16 class PickRelValInputFiles( ConfigToolBase ):
17  """ Picks up RelVal input files automatically and
18  returns a vector of strings with the paths to be used in [PoolSource].fileNames
19  PickRelValInputFiles( cmsswVersion, relVal, dataTier, condition, globalTag, maxVersions, skipFiles, numberOfFiles, debug )
20  - useDAS : switch to perform query in DAS rather than in DBS
21  optional; default: False
22  - cmsswVersion : CMSSW release to pick up the RelVal files from
23  optional; default: the current release (determined automatically from environment)
24  - formerVersion: use the last before the last valid CMSSW release to pick up the RelVal files from
25  applies also, if 'cmsswVersion' is set explicitly
26  optional; default: False
27  - relVal : RelVal sample to be used
28  optional; default: 'RelValTTbar'
29  - dataTier : data tier to be used
30  optional; default: 'GEN-SIM-RECO'
31  - condition : identifier of GlobalTag as defined in Configurations/PyReleaseValidation/python/autoCond.py
32  possibly overwritten, if 'globalTag' is set explicitly
33  optional; default: 'startup'
34  - globalTag : name of GlobalTag as it is used in the data path of the RelVals
35  optional; default: determined automatically as defined by 'condition' in Configurations/PyReleaseValidation/python/autoCond.py
36  !!! Determination is done for the release one runs in, not for the release the RelVals have been produced in.
37  !!! Example of deviation: data RelVals (CMSSW_4_1_X) might not only have the pure name of the GlobalTag 'GR_R_311_V2' in the full path,
38  but also an extension identifying the data: 'GR_R_311_V2_RelVal_wzMu2010B'
39  - maxVersions : max. versioning number of RelVal to check
40  optional; default: 9
41  - skipFiles : number of files to skip for a found RelVal sample
42  optional; default: 0
43  - numberOfFiles: number of files to pick up
44  setting it to negative values, returns all found ('skipFiles' remains active though)
45  optional; default: -1
46  - debug : switch to enable enhanced messages in 'stdout'
47  optional; default: False
48  """
49 
50  _label = 'pickRelValInputFiles'
51  _defaultParameters = dicttypes.SortedKeysDict()
52 
53  def getDefaultParameters( self ):
54  return self._defaultParameters
55 
56  def __init__( self ):
57  ConfigToolBase.__init__( self )
58  self.addParameter( self._defaultParameters, 'useDAS' , False , '' )
59  self.addParameter( self._defaultParameters, 'cmsswVersion' , os.getenv( "CMSSW_VERSION" ) , 'auto from environment' )
60  self.addParameter( self._defaultParameters, 'formerVersion', False , '' )
61  self.addParameter( self._defaultParameters, 'relVal' , 'RelValTTbar' , '' )
62  self.addParameter( self._defaultParameters, 'dataTier' , 'GEN-SIM-RECO' , '' )
63  self.addParameter( self._defaultParameters, 'condition' , 'startup' , '' )
64  gt = autoCond[ self.getDefaultParameters()[ 'condition' ].value ]
65  if isinstance(gt,tuple) or isinstance(gt,list):
66  gt = gt[0]
67  self.addParameter( self._defaultParameters, 'globalTag' , gt[ : -5 ] , 'auto from \'condition\'' )
68  self.addParameter( self._defaultParameters, 'maxVersions' , 3 , '' )
69  self.addParameter( self._defaultParameters, 'skipFiles' , 0 , '' )
70  self.addParameter( self._defaultParameters, 'numberOfFiles', -1 , 'all' )
71  self.addParameter( self._defaultParameters, 'debug' , False , '' )
72  self._parameters = copy.deepcopy( self._defaultParameters )
73  self._comment = ""
74 
75  def __call__( self
76  , useDAS = None
77  , cmsswVersion = None
78  , formerVersion = None
79  , relVal = None
80  , dataTier = None
81  , condition = None
82  , globalTag = None
83  , maxVersions = None
84  , skipFiles = None
85  , numberOfFiles = None
86  , debug = None
87  ):
88  if useDAS is None:
89  useDAS = self.getDefaultParameters()[ 'useDAS' ].value
90  if cmsswVersion is None:
91  cmsswVersion = self.getDefaultParameters()[ 'cmsswVersion' ].value
92  if formerVersion is None:
93  formerVersion = self.getDefaultParameters()[ 'formerVersion' ].value
94  if relVal is None:
95  relVal = self.getDefaultParameters()[ 'relVal' ].value
96  if dataTier is None:
97  dataTier = self.getDefaultParameters()[ 'dataTier' ].value
98  if condition is None:
99  condition = self.getDefaultParameters()[ 'condition' ].value
100  if globalTag is None:
101  globalTag = autoCond[ condition ][ : -5 ] # auto from 'condition'
102  if maxVersions is None:
103  maxVersions = self.getDefaultParameters()[ 'maxVersions' ].value
104  if skipFiles is None:
105  skipFiles = self.getDefaultParameters()[ 'skipFiles' ].value
106  if numberOfFiles is None:
107  numberOfFiles = self.getDefaultParameters()[ 'numberOfFiles' ].value
108  if debug is None:
109  debug = self.getDefaultParameters()[ 'debug' ].value
110  self.setParameter( 'useDAS' , useDAS )
111  self.setParameter( 'cmsswVersion' , cmsswVersion )
112  self.setParameter( 'formerVersion', formerVersion )
113  self.setParameter( 'relVal' , relVal )
114  self.setParameter( 'dataTier' , dataTier )
115  self.setParameter( 'condition' , condition )
116  self.setParameter( 'globalTag' , globalTag )
117  self.setParameter( 'maxVersions' , maxVersions )
118  self.setParameter( 'skipFiles' , skipFiles )
119  self.setParameter( 'numberOfFiles', numberOfFiles )
120  self.setParameter( 'debug' , debug )
121  return self.apply()
122 
123  def messageEmptyList( self ):
124  print '%s DEBUG: Empty file list returned'%( self._label )
125  print ' This might be overwritten by providing input files explicitly to the source module in the main configuration file.'
126 
127  def apply( self ):
128  useDAS = self._parameters[ 'useDAS' ].value
129  cmsswVersion = self._parameters[ 'cmsswVersion' ].value
130  formerVersion = self._parameters[ 'formerVersion' ].value
131  relVal = self._parameters[ 'relVal' ].value
132  dataTier = self._parameters[ 'dataTier' ].value
133  condition = self._parameters[ 'condition' ].value # only used for GT determination in initialization, if GT not explicitly given
134  globalTag = self._parameters[ 'globalTag' ].value
135  maxVersions = self._parameters[ 'maxVersions' ].value
136  skipFiles = self._parameters[ 'skipFiles' ].value
137  numberOfFiles = self._parameters[ 'numberOfFiles' ].value
138  debug = self._parameters[ 'debug' ].value
139 
140  filePaths = []
141 
142  # Determine corresponding CMSSW version for RelVals
143  preId = '_pre'
144  patchId = '_patch' # patch releases
145  hltPatchId = '_hltpatch' # HLT patch releases
146  dqmPatchId = '_dqmpatch' # DQM patch releases
147  slhcId = '_SLHC' # SLHC releases
148  rootId = '_root' # ROOT test releases
149  ibId = '_X_' # IBs
150  if patchId in cmsswVersion:
151  cmsswVersion = cmsswVersion.split( patchId )[ 0 ]
152  elif hltPatchId in cmsswVersion:
153  cmsswVersion = cmsswVersion.split( hltPatchId )[ 0 ]
154  elif dqmPatchId in cmsswVersion:
155  cmsswVersion = cmsswVersion.split( dqmPatchId )[ 0 ]
156  elif rootId in cmsswVersion:
157  cmsswVersion = cmsswVersion.split( rootId )[ 0 ]
158  elif slhcId in cmsswVersion:
159  cmsswVersion = cmsswVersion.split( slhcId )[ 0 ]
160  elif ibId in cmsswVersion or formerVersion:
161  outputTuple = Popen( [ 'scram', 'l -c CMSSW' ], stdout = PIPE, stderr = PIPE ).communicate()
162  if len( outputTuple[ 1 ] ) != 0:
163  print '%s INFO : SCRAM error'%( self._label )
164  if debug:
165  print ' from trying to determine last valid releases before \'%s\''%( cmsswVersion )
166  print
167  print outputTuple[ 1 ]
168  print
169  self.messageEmptyList()
170  return filePaths
171  versions = { 'last' :''
172  , 'lastToLast':''
173  }
174  for line in outputTuple[ 0 ].splitlines():
175  version = line.split()[ 1 ]
176  if cmsswVersion.split( ibId )[ 0 ] in version or cmsswVersion.rpartition( '_' )[ 0 ] in version:
177  if not ( patchId in version or hltPatchId in version or dqmPatchId in version or slhcId in version or ibId in version or rootId in version ):
178  versions[ 'lastToLast' ] = versions[ 'last' ]
179  versions[ 'last' ] = version
180  if version == cmsswVersion:
181  break
182  # FIXME: ordering of output problematic ('XYZ_pre10' before 'XYZ_pre2', no "formerVersion" for 'XYZ_pre1')
183  if formerVersion:
184  # Don't use pre-releases as "former version" for other releases than CMSSW_X_Y_0
185  if preId in versions[ 'lastToLast' ] and not preId in versions[ 'last' ] and not versions[ 'last' ].endswith( '_0' ):
186  versions[ 'lastToLast' ] = versions[ 'lastToLast' ].split( preId )[ 0 ] # works only, if 'CMSSW_X_Y_0' esists ;-)
187  # Use pre-release as "former version" for CMSSW_X_Y_0
188  elif versions[ 'last' ].endswith( '_0' ) and not ( preId in versions[ 'lastToLast' ] and versions[ 'lastToLast' ].startswith( versions[ 'last' ] ) ):
189  versions[ 'lastToLast' ] = ''
190  for line in outputTuple[ 0 ].splitlines():
191  version = line.split()[ 1 ]
192  versionParts = version.partition( preId )
193  if versionParts[ 0 ] == versions[ 'last' ] and versionParts[ 1 ] == preId:
194  versions[ 'lastToLast' ] = version
195  elif versions[ 'lastToLast' ] != '':
196  break
197  # Don't use CMSSW_X_Y_0 as "former version" for pre-releases
198  elif preId in versions[ 'last' ] and not preId in versions[ 'lastToLast' ] and versions[ 'lastToLast' ].endswith( '_0' ):
199  versions[ 'lastToLast' ] = '' # no alternative :-(
200  cmsswVersion = versions[ 'lastToLast' ]
201  else:
202  cmsswVersion = versions[ 'last' ]
203 
204  # Debugging output
205  if debug:
206  print '%s DEBUG: Called with...'%( self._label )
207  for key in self._parameters.keys():
208  print ' %s:\t'%( key ),
209  print self._parameters[ key ].value,
210  if self._parameters[ key ].value is self.getDefaultParameters()[ key ].value:
211  print ' (default)'
212  else:
213  print
214  if key == 'cmsswVersion' and cmsswVersion != self._parameters[ key ].value:
215  if formerVersion:
216  print ' ==> modified to last to last valid release %s (s. \'formerVersion\' parameter)'%( cmsswVersion )
217  else:
218  print ' ==> modified to last valid release %s'%( cmsswVersion )
219 
220  # Check domain
221  domain = socket.getfqdn().split( '.' )
222  domainSE = ''
223  if len( domain ) == 0:
224  print '%s INFO : Cannot determine domain of this computer'%( self._label )
225  if debug:
226  self.messageEmptyList()
227  return filePaths
228  elif os.uname()[0] == "Darwin":
229  print '%s INFO : Running on MacOSX without direct access to RelVal files.'%( self._label )
230  if debug:
231  self.messageEmptyList()
232  return filePaths
233  elif len( domain ) == 1:
234  print '%s INFO : Running on local host \'%s\' without direct access to RelVal files'%( self._label, domain[ 0 ] )
235  if debug:
236  self.messageEmptyList()
237  return filePaths
238  if not ( ( domain[ -2 ] == 'cern' and domain[ -1 ] == 'ch' ) or ( domain[ -2 ] == 'fnal' and domain[ -1 ] == 'gov' ) ):
239  print '%s INFO : Running on site \'%s.%s\' without direct access to RelVal files'%( self._label, domain[ -2 ], domain[ -1 ] )
240  if debug:
241  self.messageEmptyList()
242  return filePaths
243  if domain[ -2 ] == 'cern':
244  domainSE = 'T2_CH_CERN'
245  elif domain[ -2 ] == 'fnal':
246  domainSE = 'T1_US_FNAL_MSS'
247  if debug:
248  print '%s DEBUG: Running at site \'%s.%s\''%( self._label, domain[ -2 ], domain[ -1 ] )
249  print '%s DEBUG: Looking for SE \'%s\''%( self._label, domainSE )
250 
251  # Find files
252  validVersion = 0
253  dataset = ''
254  datasetAll = '/%s/%s-%s-v*/%s'%( relVal, cmsswVersion, globalTag, dataTier )
255  if useDAS:
256  if debug:
257  print '%s DEBUG: Using DAS query'%( self._label )
258  dasLimit = numberOfFiles
259  if dasLimit <= 0:
260  dasLimit = 1
261  for version in range( maxVersions, 0, -1 ):
262  filePaths = []
263  filePathsTmp = []
264  fileCount = 0
265  dataset = '/%s/%s-%s-v%i/%s'%( relVal, cmsswVersion, globalTag, version, dataTier )
266  dasQuery = 'file dataset=%s | grep file.name'%( dataset )
267  if debug:
268  print '%s DEBUG: Querying dataset \'%s\' with'%( self._label, dataset )
269  print ' \'%s\''%( dasQuery )
270  jsondict = das_client.get_data(dasQuery,dasLimit)
271  if debug:
272  print '%s DEBUG: Received DAS JSON dictionary:'%( self._label )
273  print ' \'%s\''%( jsondict )
274  if jsondict[ 'status' ] != 'ok':
275  print 'There was a problem while querying DAS with query \'%s\'. Server reply was:\n %s' % (dasQuery, jsondict)
276  exit( 1 )
277  mongo_query = jsondict[ 'mongo_query' ]
278  filters = mongo_query[ 'filters' ]
279  data = jsondict[ 'data' ]
280  if debug:
281  print '%s DEBUG: Query in JSON dictionary:'%( self._label )
282  print ' \'%s\''%( mongo_query )
283  print '%s DEBUG: Filters in query:'%( self._label )
284  print ' \'%s\''%( filters )
285  print '%s DEBUG: Data in JSON dictionary:'%( self._label )
286  print ' \'%s\''%( data )
287  for row in data:
288  filePath = [ r for r in das_client.get_value( row, filters[ 'grep' ] ) ][ 0 ]
289  if debug:
290  print '%s DEBUG: Testing file entry \'%s\''%( self._label, filePath )
291  if len( filePath ) > 0:
292  if validVersion != version:
293  jsontestdict = das_client.get_data('site dataset=%s | grep site.name' % ( dataset ), 999)
294  mongo_testquery = jsontestdict[ 'mongo_query' ]
295  testfilters = mongo_testquery[ 'filters' ]
296  testdata = jsontestdict[ 'data' ]
297  if debug:
298  print '%s DEBUG: Received DAS JSON dictionary (site test):'%( self._label )
299  print ' \'%s\''%( jsontestdict )
300  print '%s DEBUG: Query in JSON dictionary (site test):'%( self._label )
301  print ' \'%s\''%( mongo_testquery )
302  print '%s DEBUG: Filters in query (site test):'%( self._label )
303  print ' \'%s\''%( testfilters )
304  print '%s DEBUG: Data in JSON dictionary (site test):'%( self._label )
305  print ' \'%s\''%( testdata )
306  foundSE = False
307  for testrow in testdata:
308  siteName = [ tr for tr in das_client.get_value( testrow, testfilters[ 'grep' ] ) ][ 0 ]
309  if siteName == domainSE:
310  foundSE = True
311  break
312  if not foundSE:
313  if debug:
314  print '%s DEBUG: Possible version \'v%s\' not available on SE \'%s\''%( self._label, version, domainSE )
315  break
316  validVersion = version
317  if debug:
318  print '%s DEBUG: Valid version set to \'v%i\''%( self._label, validVersion )
319  if numberOfFiles == 0:
320  break
321  # protect from double entries ( 'unique' flag in query does not work here)
322  if not filePath in filePathsTmp:
323  filePathsTmp.append( filePath )
324  if debug:
325  print '%s DEBUG: File \'%s\' found'%( self._label, filePath )
326  fileCount += 1
327  # needed, since and "limit" overrides "idx" in 'get_data' (==> "idx" set to '0' rather than "skipFiles")
328  if fileCount > skipFiles:
329  filePaths.append( filePath )
330  elif debug:
331  print '%s DEBUG: File \'%s\' found again'%( self._label, filePath )
332  if validVersion > 0:
333  if numberOfFiles == 0 and debug:
334  print '%s DEBUG: No files requested'%( self._label )
335  break
336  else:
337  if debug:
338  print '%s DEBUG: Using DBS query'%( self._label )
339  print '%s WARNING: DBS query disabled for DBS3 transition to new API'%( self._label )
340  #for version in range( maxVersions, 0, -1 ):
341  #filePaths = []
342  #fileCount = 0
343  #dataset = '/%s/%s-%s-v%i/%s'%( relVal, cmsswVersion, globalTag, version, dataTier )
344  #dbsQuery = 'find file where dataset = %s'%( dataset )
345  #if debug:
346  #print '%s DEBUG: Querying dataset \'%s\' with'%( self._label, dataset )
347  #print ' \'%s\''%( dbsQuery )
348  #foundSE = False
349  #for line in os.popen( 'dbs search --query="%s"'%( dbsQuery ) ).readlines():
350  #if line.find( '.root' ) != -1:
351  #if validVersion != version:
352  #if not foundSE:
353  #dbsSiteQuery = 'find dataset where dataset = %s and site = %s'%( dataset, domainSE )
354  #if debug:
355  #print '%s DEBUG: Querying site \'%s\' with'%( self._label, domainSE )
356  #print ' \'%s\''%( dbsSiteQuery )
357  #for lineSite in os.popen( 'dbs search --query="%s"'%( dbsSiteQuery ) ).readlines():
358  #if lineSite.find( dataset ) != -1:
359  #foundSE = True
360  #break
361  #if not foundSE:
362  #if debug:
363  #print '%s DEBUG: Possible version \'v%s\' not available on SE \'%s\''%( self._label, version, domainSE )
364  #break
365  #validVersion = version
366  #if debug:
367  #print '%s DEBUG: Valid version set to \'v%i\''%( self._label, validVersion )
368  #if numberOfFiles == 0:
369  #break
370  #filePath = line.replace( '\n', '' )
371  #if debug:
372  #print '%s DEBUG: File \'%s\' found'%( self._label, filePath )
373  #fileCount += 1
374  #if fileCount > skipFiles:
375  #filePaths.append( filePath )
376  #if not numberOfFiles < 0:
377  #if numberOfFiles <= len( filePaths ):
378  #break
379  #if validVersion > 0:
380  #if numberOfFiles == 0 and debug:
381  #print '%s DEBUG: No files requested'%( self._label )
382  #break
383 
384  # Check output and return
385  if validVersion == 0:
386  print '%s WARNING : No RelVal file(s) found at all in datasets \'%s*\' on SE \'%s\''%( self._label, datasetAll, domainSE )
387  if debug:
388  self.messageEmptyList()
389  elif len( filePaths ) == 0:
390  print '%s WARNING : No RelVal file(s) picked up in dataset \'%s\''%( self._label, dataset )
391  if debug:
392  self.messageEmptyList()
393  elif len( filePaths ) < numberOfFiles:
394  print '%s INFO : Only %i RelVal file(s) instead of %i picked up in dataset \'%s\''%( self._label, len( filePaths ), numberOfFiles, dataset )
395 
396  if debug:
397  print '%s DEBUG: returning %i file(s):\n%s'%( self._label, len( filePaths ), filePaths )
398  return filePaths
399 
400 pickRelValInputFiles = PickRelValInputFiles()
communicate
static void * communicate(void *obj)
Definition: DQMNet.cc:1251
cmsswVersionTools.PickRelValInputFiles.__call__
def __call__(self, useDAS=None, cmsswVersion=None, formerVersion=None, relVal=None, dataTier=None, condition=None, globalTag=None, maxVersions=None, skipFiles=None, numberOfFiles=None, debug=None)
Definition: cmsswVersionTools.py:87
cmsswVersionTools.PickRelValInputFiles
Automatic pick-up of RelVal input files
Definition: cmsswVersionTools.py:16
split
double split
Definition: MVATrainer.cc:139
Generated for CMSSW Reference Manual by doxygen 1.8.11