CMS 3D CMS Logo
 All Classes Namespaces Files Functions Variables Typedefs Enumerations Enumerator Properties Friends Macros Pages
List of all members | Public Member Functions | Public Attributes
edmIntegrityCheck.IntegrityCheck Class Reference
Inheritance diagram for edmIntegrityCheck.IntegrityCheck:

Public Member Functions

def __init__
 
def getParseNumberOfEvents
 
def listFiles
 
def listRootFiles
 
def query
 
def report
 
def sortByBaseDir
 
def stageHost
 
def stripDuplicates
 
def structured
 
def test
 
def testFile
 
def testFileTimeOut
 

Public Attributes

 dataset
 
 directory
 
 eventsSeen
 
 eventsTotal
 
 options
 
 test_result
 
 topdir
 

Detailed Description

Definition at line 71 of file edmIntegrityCheck.py.

Constructor & Destructor Documentation

def edmIntegrityCheck.IntegrityCheck.__init__ (   self,
  dataset,
  options 
)

Definition at line 73 of file edmIntegrityCheck.py.

73 
74  def __init__(self, dataset, options):
75  if not dataset.startswith(os.sep):
76  dataset = os.sep + dataset
77 
78  self.dataset = dataset
79  self.options = options
80  self.topdir = castortools.lfnToCastor( castorBaseDir(user=options.user) )
81  self.directory = os.path.join(self.topdir, *self.dataset.split(os.sep))
82 
83  #event counters
84  self.eventsTotal = -1
85  self.eventsSeen = 0
86 
87  self.test_result = None

Member Function Documentation

def edmIntegrityCheck.IntegrityCheck.getParseNumberOfEvents (   self,
  output 
) 
Parse the output of edmFileUtil to get the number of events found

Definition at line 312 of file edmIntegrityCheck.py.

Referenced by edmIntegrityCheck.IntegrityCheck.testFile().

312 
313  def getParseNumberOfEvents(self,output):
314  """Parse the output of edmFileUtil to get the number of events found"""
315  tokens = output.split(' ')
316  result = -2
317  try:
318  result = int(tokens[-4])
319  except ValueError:
320  pass
321  return result
def edmIntegrityCheck.IntegrityCheck.listFiles (   self,
  dir 
) 
Recursively list a file or directory on castor

Definition at line 293 of file edmIntegrityCheck.py.

Referenced by edmIntegrityCheck.IntegrityCheck.listRootFiles().

293 
294  def listFiles(self,dir):
295  """Recursively list a file or directory on castor"""
296  return castortools.listFiles(dir,self.options.resursive)
def edmIntegrityCheck.IntegrityCheck.listRootFiles (   self,
  dir 
) 
filter out filenames so that they only contain root files

Definition at line 297 of file edmIntegrityCheck.py.

References edmIntegrityCheck.IntegrityCheck.listFiles().

Referenced by edmIntegrityCheck.IntegrityCheck.test().

297 
298  def listRootFiles(self,dir):
299  """filter out filenames so that they only contain root files"""
300  return [f for f in self.listFiles(dir) if f.lower().endswith('.root')]
def edmIntegrityCheck.IntegrityCheck.query (   self) 
Query DAS to find out how many events are in the dataset

Definition at line 88 of file edmIntegrityCheck.py.

References CalibratedPatElectronProducer.dataset, CalibratedElectronProducer.dataset, edmIntegrityCheck.IntegrityCheck.dataset, genericValidation.GenericValidationData.dataset, edmIntegrityCheck.IntegrityCheck.eventsTotal, FileExportPlugin.FileExportPlugin.options, cmsswPreprocessor.CmsswPreprocessor.options, DOTExport.DotProducer.options, confdb.HLTProcess.options, and edmIntegrityCheck.IntegrityCheck.options.

Referenced by production_tasks.BaseDataset.run(), and edmIntegrityCheck.IntegrityCheck.test().

88 
89  def query(self):
90  """Query DAS to find out how many events are in the dataset"""
91  from production_tasks import BaseDataset
92  base = BaseDataset(self.dataset, self.options.user, self.options)
93 
94  data = None
95  output = base.run({})
96  if output.has_key('Das'):
97  self.options.name = output['Name']
98  data = output['Das']
99 
100  if data is None:
101  raise Exception("Dataset '%s' not found in Das. Please check." % self.dataset)
102  #get the number of events in the dataset
103  self.eventsTotal = CMSDataset.findPrimaryDatasetEntries(self.options.name, self.options.min_run, self.options.max_run)
def edmIntegrityCheck.IntegrityCheck.report (   self)

Definition at line 211 of file edmIntegrityCheck.py.

References pat::GenericDuplicateRemover< Comparator, Arbitrator >.duplicates(), LumiList.LumiList.duplicates, OniaPhotonConversionProducer.duplicates, edmIntegrityCheck.IntegrityCheck.eventsSeen, edmIntegrityCheck.IntegrityCheck.eventsTotal, join(), value_test.ValueTestCase.test(), eventstfile_test.EventsTFileTestCase.test(), pat::Flags.test(), helper::Parser.test(), reco::PFBlock::Link.test, pftools::CaloBox.test(), helper::ScannerBase.test(), edmIntegrityCheck.IntegrityCheck.test(), XMLProcessor.test(), DTTFBitArray< N >.test(), BitArray< N >.test(), TwoObjectVariable< LHS, lLHS, RHS, lRHS, Calculator >::getObject.test, cond::BlobStreamingService.test::BlobStreaming::test, reco::HitPattern.test::TestHitPattern::test, edmIntegrityCheck.IntegrityCheck.test_result, and edmIntegrityCheck.IntegrityCheck.topdir.

211 
212  def report(self):
213 
214  if self.test_result is None:
215  self.test()
216 
217  print 'DBS Dataset name: %s' % self.options.name
218  print 'Storage path: %s' % self.topdir
219 
220  for dirname, files in self.test_result.iteritems():
221  print 'Directory: %s' % dirname
222  for name, status in files.iteritems():
223  fname = os.path.join(dirname, name)
224  if not fname in self.duplicates:
225  print '\t\t %s: %s' % (name, str(status))
226  else:
227  print '\t\t %s: %s (Valid duplicate)' % (name, str(status))
228  print 'Total entries in DBS: %i' % self.eventsTotal
229  print 'Total entries in processed files: %i' % self.eventsSeen
230  if self.eventsTotal>0:
231  print 'Fraction of dataset processed: %f' % (self.eventsSeen/(1.*self.eventsTotal))
232  else:
233  print 'Total entries in DBS not determined'
234  if self.bad_jobs:
235  print "Bad Crab Jobs: '%s'" % ','.join([str(j) for j in self.bad_jobs])
join
static std::string join(char **cmd)
Definition: RemoteFile.cc:18
def edmIntegrityCheck.IntegrityCheck.sortByBaseDir (   self,
  files 
) 
Sort files into directories

Definition at line 301 of file edmIntegrityCheck.py.

References python.multivaluedict.append().

Referenced by edmIntegrityCheck.IntegrityCheck.test().

301 
302  def sortByBaseDir(self,files):
303  """Sort files into directories"""
304  result = {}
305  for f in files:
306  dirname = os.path.dirname(f)
307  filename = os.path.basename(f)
308  if not result.has_key(dirname): result[dirname] = []
309  result[dirname].append(filename)
310  return result
311 
def edmIntegrityCheck.IntegrityCheck.stageHost (   self) 
Returns the CASTOR instance to use

Definition at line 289 of file edmIntegrityCheck.py.

Referenced by edmIntegrityCheck.IntegrityCheck.structured().

289 
290  def stageHost(self):
291  """Returns the CASTOR instance to use"""
292  return os.environ.get('STAGE_HOST','castorcms')
def edmIntegrityCheck.IntegrityCheck.stripDuplicates (   self)

Definition at line 104 of file edmIntegrityCheck.py.

References python.multivaluedict.append(), list(), bookConverter.max, and min().

Referenced by edmIntegrityCheck.IntegrityCheck.test().

104 
105  def stripDuplicates(self):
106 
107  import re
108 
109  filemask = {}
110  for dirname, files in self.test_result.iteritems():
111  for name, status in files.iteritems():
112  fname = os.path.join(dirname, name)
113  filemask[fname] = status
114 
115  def isCrabFile(name):
116  _, fname = os.path.split(name)
117  base, _ = os.path.splitext(fname)
118  return re.match(".*_\d+_\d+_\w+$", base) is not None, base
119  def getCrabIndex(base):
120  tokens = base.split('_')
121  if len(tokens) > 2:
122  return (int(tokens[-3]), int(tokens[-2]))
123  return None
124 
125  files = {}
126 
127  mmin = 1000000000
128  mmax = -100000000
129  for f in filemask:
130  isCrab, base = isCrabFile(f)
131  if isCrab:
132  index = getCrabIndex(base)
133  if index is not None:
134  jobid, retry = index
135 
136  mmin = min(mmin, jobid)
137  mmax = max(mmax, jobid)
138  if files.has_key(jobid) and filemask[f][0]:
139  files[jobid].append((retry, f))
140  elif filemask[f][0]:
141  files[jobid] = [(retry, f)]
142 
143  good_duplicates = {}
144  bad_jobs = set()
145  sum_dup = 0
146  for i in xrange(mmin, mmax+1):
147  if files.has_key(i):
148  duplicates = files[i]
149  duplicates.sort()
150 
151  fname = duplicates[-1][1]
152  if len(duplicates) > 1:
153  for d in duplicates[:-1]:
154  good_duplicates[d[1]] = filemask[d[1]][1]
155  sum_dup += good_duplicates[d[1]]
156  else:
157  bad_jobs.add(i)
158  return good_duplicates, sorted(list(bad_jobs)), sum_dup
min
T min(T a, T b)
Definition: MathUtil.h:58
list
How EventSelector::AcceptEvent() decides whether to accept an event for output otherwise it is excluding the probing of A single or multiple positive and the trigger will pass if any such matching triggers are PASS or EXCEPTION[A criterion thatmatches no triggers at all is detected and causes a throw.] A single negative with an expectation of appropriate bit checking in the decision and the trigger will pass if any such matching triggers are FAIL or EXCEPTION A wildcarded negative criterion that matches more than one trigger in the trigger list("!*","!HLTx*"if it matches 2 triggers or more) will accept the event if all the matching triggers are FAIL.It will reject the event if any of the triggers are PASS or EXCEPTION(this matches the behavior of"!*"before the partial wildcard feature was incorporated).Triggers which are in the READY state are completely ignored.(READY should never be returned since the trigger paths have been run
def edmIntegrityCheck.IntegrityCheck.structured (   self)

Definition at line 236 of file edmIntegrityCheck.py.

References python.multivaluedict.append(), CalibratedPatElectronProducer.dataset, CalibratedElectronProducer.dataset, edmIntegrityCheck.IntegrityCheck.dataset, genericValidation.GenericValidationData.dataset, pat::GenericDuplicateRemover< Comparator, Arbitrator >.duplicates(), LumiList.LumiList.duplicates, OniaPhotonConversionProducer.duplicates, edmIntegrityCheck.IntegrityCheck.eventsSeen, edmIntegrityCheck.IntegrityCheck.eventsTotal, edmIntegrityCheck.IntegrityCheck.stageHost(), value_test.ValueTestCase.test(), eventstfile_test.EventsTFileTestCase.test(), pat::Flags.test(), helper::Parser.test(), pftools::CaloBox.test(), reco::PFBlock::Link.test, helper::ScannerBase.test(), edmIntegrityCheck.IntegrityCheck.test(), XMLProcessor.test(), BitArray< N >.test(), DTTFBitArray< N >.test(), TwoObjectVariable< LHS, lLHS, RHS, lRHS, Calculator >::getObject.test, cond::BlobStreamingService.test::BlobStreaming::test, reco::HitPattern.test::TestHitPattern::test, edmIntegrityCheck.IntegrityCheck.test_result, and edmIntegrityCheck.IntegrityCheck.topdir.

236 
237  def structured(self):
238 
239  if self.test_result is None:
240  self.test()
241 
242  totalGood = 0
243  totalBad = 0
244 
245  report = {'data':{},
246  'ReportVersion':3,
247  'PrimaryDataset':self.options.name,
248  'Name':self.dataset,
249  'PhysicsGroup':'CMG',
250  'Status':'VALID',
251  'TierList':[],
252  'AlgoList':[],
253  'RunList':[],
254  'PathList':[],
255  'Topdir':self.topdir,
256  'StageHost':self.stageHost(),
257  'CreatedBy':self.options.user,
258  'DateCreated':datetime.datetime.now().strftime("%s"),
259  'Files':{}}
260 
261  for dirname, files in self.test_result.iteritems():
262  report['PathList'].append(dirname)
263  for name, status in files.iteritems():
264  fname = os.path.join(dirname, name)
265  report['Files'][fname] = status
266  if status[0]:
267  totalGood += 1
268  else:
269  totalBad += 1
270 
271  report['PrimaryDatasetEntries'] = self.eventsTotal
272  if self.eventsTotal>0:
273  report['PrimaryDatasetFraction'] = (self.eventsSeen/(1.*self.eventsTotal))
274  else:
275  report['PrimaryDatasetFraction'] = -1.
276  report['FilesEntries'] = self.eventsSeen
277 
278  report['FilesGood'] = totalGood
279  report['FilesBad'] = totalBad
280  report['FilesCount'] = totalGood + totalBad
281 
282  report['BadJobs'] = self.bad_jobs
283  report['ValidDuplicates'] = self.duplicates
284 
285  report['MinRun'] = self.options.min_run
286  report['MaxRun'] = self.options.max_run
287 
288  return report
def edmIntegrityCheck.IntegrityCheck.test (   self,
  previous = None,
  timeout = -1 
)

Definition at line 159 of file edmIntegrityCheck.py.

References Book.directory, edmIntegrityCheck.IntegrityCheck.directory, pat::GenericDuplicateRemover< Comparator, Arbitrator >.duplicates(), LumiList.LumiList.duplicates, OniaPhotonConversionProducer.duplicates, edmIntegrityCheck.IntegrityCheck.eventsSeen, edmIntegrityCheck.IntegrityCheck.listRootFiles(), DbQuery.query, edmIntegrityCheck.IntegrityCheck.query(), confdbOfflineConverter.OfflineConverter.query(), ora::QueryableVector< Tp >.query(), upload_popcon.HTTP.query(), edmIntegrityCheck.IntegrityCheck.sortByBaseDir(), edmIntegrityCheck.IntegrityCheck.stripDuplicates(), edmIntegrityCheck.IntegrityCheck.test_result, and edmIntegrityCheck.IntegrityCheck.testFileTimeOut().

Referenced by edmIntegrityCheck.IntegrityCheck.report(), and edmIntegrityCheck.IntegrityCheck.structured().

159 
160  def test(self, previous = None, timeout = -1):
161  if not castortools.fileExists(self.directory):
162  raise Exception("The top level directory '%s' for this dataset does not exist" % self.directory)
163 
164  self.query()
165 
166  test_results = {}
167 
168  #support updating to speed things up
169  prev_results = {}
170  if previous is not None:
171  for name, status in previous['Files'].iteritems():
172  prev_results[name] = status
173 
174  filesToTest = self.sortByBaseDir(self.listRootFiles(self.directory))
175  for dir, filelist in filesToTest.iteritems():
176  filemask = {}
177  #apply a UNIX wildcard if specified
178  filtered = filelist
179  if self.options.wildcard is not None:
180  filtered = fnmatch.filter(filelist, self.options.wildcard)
181  if not filtered:
182  print >> sys.stderr, "Warning: The wildcard '%s' does not match any files in '%s'. Please check you are using quotes." % (self.options.wildcard,self.directory)
183 
184  count = 0
185  for ff in filtered:
186  fname = os.path.join(dir, ff)
187  lfn = castortools.castorToLFN(fname)
188 
189  #try to update from the previous result if available
190  if lfn in prev_results and prev_results[lfn][0]:
191  if self.options.printout:
192  print '[%i/%i]\t Skipping %s...' % (count, len(filtered),fname),
193  OK, num = prev_results[lfn]
194  else:
195  if self.options.printout:
196  print '[%i/%i]\t Checking %s...' % (count, len(filtered),fname),
197  OK, num = self.testFileTimeOut(lfn, timeout)
198 
199  filemask[ff] = (OK,num)
200  if self.options.printout:
201  print (OK, num)
202  if OK:
203  self.eventsSeen += num
204  count += 1
205  test_results[castortools.castorToLFN(dir)] = filemask
206  self.test_result = test_results
207 
208  self.duplicates, self.bad_jobs, sum_dup = self.stripDuplicates()
209  #remove duplicate entries from the event count
210  self.eventsSeen -= sum_dup
def edmIntegrityCheck.IntegrityCheck.testFile (   self,
  lfn 
)

Definition at line 322 of file edmIntegrityCheck.py.

References communicate(), and edmIntegrityCheck.IntegrityCheck.getParseNumberOfEvents().

Referenced by edmIntegrityCheck.IntegrityCheck.testFileTimeOut().

322 
323  def testFile(self,lfn):
324  stdout = subprocess.Popen(['edmFileUtil',lfn], stdout=subprocess.PIPE,stderr=subprocess.PIPE).communicate()[0]
325  for error in ["Fatal Root Error","Could not open file","Not a valid collection"]:
326  if error in stdout: return (False,-1)
327  return (True,self.getParseNumberOfEvents(stdout))
communicate
static void * communicate(void *obj)
Definition: DQMNet.cc:1246
def edmIntegrityCheck.IntegrityCheck.testFileTimeOut (   self,
  lfn,
  timeout 
)

Definition at line 328 of file edmIntegrityCheck.py.

References AlignmentIORootBase.testFile(), edmIntegrityCheck.IntegrityCheck.testFile(), and timeout.timed_out().

Referenced by edmIntegrityCheck.IntegrityCheck.test().

328 
329  def testFileTimeOut(self,lfn, timeout):
330  @timed_out(timeout)
331  def tf(lfn):
332  try:
333  return self.testFile(lfn)
334  except TimedOutExc, e:
335  print >> sys.stderr, "ERROR:\tedmFileUtil timed out for lfn '%s' (%d)" % (lfn,timeout)
336  return (False,-1)
337  if timeout > 0:
338  return tf(lfn)
339  else:
340  return self.testFile(lfn)
341 
342 
timeout.timed_out
def timed_out
Definition: timeout.py:23

Member Data Documentation

edmIntegrityCheck.IntegrityCheck.dataset

Definition at line 77 of file edmIntegrityCheck.py.

Referenced by edmIntegrityCheck.IntegrityCheck.query(), production_tasks.BaseDataset.query(), production_tasks.CheckDatasetExists.run(), production_tasks.BaseDataset.run(), production_tasks.GenerateMask.run(), production_tasks.CreateJobDirectory.run(), production_tasks.SourceCFG.run(), production_tasks.FullCFG.run(), production_tasks.WriteToDatasets.run(), production_tasks.RunCMSBatch.run(), production_tasks.MonitorJobs.run(), production_tasks.CleanJobFiles.run(), and edmIntegrityCheck.IntegrityCheck.structured().

edmIntegrityCheck.IntegrityCheck.directory

Definition at line 80 of file edmIntegrityCheck.py.

Referenced by dqm_interfaces.DirFetcher.run(), utils.unpickler.run(), edmIntegrityCheck.IntegrityCheck.test(), and dqm_interfaces.DirWalkerFile.walk().

edmIntegrityCheck.IntegrityCheck.eventsSeen

Definition at line 84 of file edmIntegrityCheck.py.

Referenced by edmIntegrityCheck.IntegrityCheck.report(), edmIntegrityCheck.IntegrityCheck.structured(), and edmIntegrityCheck.IntegrityCheck.test().

edmIntegrityCheck.IntegrityCheck.eventsTotal

Definition at line 83 of file edmIntegrityCheck.py.

Referenced by edmIntegrityCheck.IntegrityCheck.query(), edmIntegrityCheck.IntegrityCheck.report(), and edmIntegrityCheck.IntegrityCheck.structured().

edmIntegrityCheck.IntegrityCheck.options

Definition at line 78 of file edmIntegrityCheck.py.

Referenced by betterConfigParser.BetterConfigParser.__updateDict(), betterConfigParser.BetterConfigParser.checkInput(), DOTExport.DotExport.export(), edmIntegrityCheck.IntegrityCheck.query(), production_tasks.BaseDataset.run(), production_tasks.GenerateMask.run(), production_tasks.RunCMSBatch.run(), production_tasks.MonitorJobs.run(), production_tasks.CheckJobStatus.run(), production_tasks.CleanJobFiles.run(), and DOTExport.DotExport.write_output().

edmIntegrityCheck.IntegrityCheck.test_result

Definition at line 86 of file edmIntegrityCheck.py.

Referenced by edmIntegrityCheck.IntegrityCheck.report(), edmIntegrityCheck.IntegrityCheck.structured(), and edmIntegrityCheck.IntegrityCheck.test().

edmIntegrityCheck.IntegrityCheck.topdir

Definition at line 79 of file edmIntegrityCheck.py.

Referenced by edmIntegrityCheck.IntegrityCheck.report(), and edmIntegrityCheck.IntegrityCheck.structured().

Generated for CMSSW Reference Manual by doxygen 1.8.5