CMS 3D CMS Logo

 All Classes Namespaces Files Functions Variables Typedefs Enumerations Enumerator Properties Friends Macros Pages
Functions | Variables
ValidationMatrix Namespace Reference

Functions

def add_to_blacklist
 
def call_compare_using_files
 
def check_root_files
 
def count_alive_processes
 
def do_comparisons_threaded
 
def do_html
 
def do_reports
 
def get_clean_fileanames
 
def get_filenames_from_pool
 
def get_roofiles_in_dir
 
def guess_blacklists
 
def guess_params
 
def name2globaltag
 
def name2run
 
def name2runskim
 
def name2sample
 
def name2version
 

Variables

string action = "store"
 —HASHING—## More...
 
string all_samples = ""
 
 argv = sys.argv
 
 default = ref_samples,
 
string dest = "ref_samples"
 
string help = "The samples that act as reference (comma separated list)"
 
 hlt = False
 
string in_dir = ""
 
int n_processes = 1
 
int n_threads = 1
 
string out_dir = ""
 
tuple parser = OptionParser(usage="usage: %prog [options]")
 
string ref_samples = ""
 
int run = -1
 
string stat_test = "Chi2"
 
string test_samples = ""
 
float test_threshold = 0.00001
 

Function Documentation

def ValidationMatrix.add_to_blacklist (   blacklist,
  pattern,
  target,
  blist_piece 
)

Definition at line 125 of file ValidationMatrix.py.

References cond::persistency.search().

Referenced by guess_blacklists().

126 def add_to_blacklist(blacklist, pattern, target, blist_piece):
127  int_pattern=pattern
128  int_pattern=pattern.strip()
129  flip_condition=False
130  if int_pattern[0]=='!':
131  int_pattern=int_pattern[1:]
132  flip_condition=True
133 
134  condition = search(int_pattern,target)!=None
135  if flip_condition:
136  condition = not condition
137 
138  if condition:
139  #print "Found %s in %s" %(pattern,target)
140  if blacklist!="": # if not the first, add a comma
141  blacklist+=","
142  blacklist+=blist_piece
143  #else:
144  #print " NOT Found %s in %s" %(pattern,target)
145  return blacklist
146 
147 #-------------------------------------------------------------------------------
std::vector< T >::const_iterator search(const cond::Time_t &val, const std::vector< T > &container)
Definition: IOVProxy.cc:252
def ValidationMatrix.call_compare_using_files (   args)
Creates shell command to compare two files using compare_using_files.py
script and calls it.

Definition at line 272 of file ValidationMatrix.py.

References alcazmumu_cfi.filter, guess_blacklists(), name2globaltag(), and name2version().

273 def call_compare_using_files(args):
274  """Creates shell command to compare two files using compare_using_files.py
275  script and calls it."""
276  sample, ref_filename, test_filename, options = args
277  gt = name2globaltag(ref_filename)
278  blacklists=guess_blacklists([sample],name2version(ref_filename),name2version(test_filename),options.hlt)
279  command = " compare_using_files.py "
280  command+= "%s %s " %(ref_filename,test_filename)
281  command+= " -C -R "
282  if options.do_pngs:
283  command+= " -p "
284  command+= " -o %s_%s " %(sample, gt)
285  # Change threshold to an experimental and empirical value of 10^-5
286  command+= " --specify_run "
287  if options.stat_test in ["Bin2Bin", "BinToBin"]:
288  options.test_threshold = 0.9999
289  command+= " -t %s " %options.test_threshold
290  command+= " -s %s " %options.stat_test
291 
292  # Inspect the HLT directories
293  if options.hlt:
294  command+=" -d HLT "
295 
296  if options.hash_name:
297  command += " --hash_name "
298 
299  if options.blacklist_file:
300  command += " --use_black_file "
301 
302  if options.standalone:
303  command += " --standalone "
304  if len(blacklists[sample]) >0:
305  command+= '-B %s ' %blacklists[sample]
306  print "\nExecuting -- %s" %command
307 
308  process=call(filter(lambda x: len(x)>0,command.split(" ")))
309  return process
310 
311 
312 #--------------------------------------------------------------------------------
def ValidationMatrix.check_root_files (   names_list)

Definition at line 116 of file ValidationMatrix.py.

Referenced by get_clean_fileanames().

117 def check_root_files(names_list):
118  for name in names_list:
119  if not name.endswith(".root"):
120  print "File %s does not seem to be a rootfile. Please check."
121  return False
122  return True
123 
124 #-------------------------------------------------------------------------------
def ValidationMatrix.count_alive_processes (   p_list)

Definition at line 267 of file ValidationMatrix.py.

References alcazmumu_cfi.filter.

268 def count_alive_processes(p_list):
269  return len(filter(lambda p: p.returncode==None,p_list))
270 
271 #-------------------------------------------------------------------------------
def ValidationMatrix.do_comparisons_threaded (   options)

Definition at line 313 of file ValidationMatrix.py.

References get_clean_fileanames(), get_filenames_from_pool(), guess_params(), list(), python.multivaluedict.map(), cond::persistency.search(), and archive.zip.

314 def do_comparisons_threaded(options):
315 
316  n_processes= int(options.n_processes)
317 
318  ref_filenames=[]
319  test_filenames=[]
320 
321  if len(options.all_samples)>0:
322  ref_filenames,test_filenames=get_filenames_from_pool(options.all_samples)
323  else:
324  ref_filenames,test_filenames=get_clean_fileanames(options.ref_samples,options.test_samples)
325 
326  # make the paths absolute
327  ref_filenames=map(os.path.abspath,ref_filenames)
328  test_filenames=map(os.path.abspath,test_filenames)
329 
330  samples,cmssw_version1,cmssw_version2=guess_params(ref_filenames,test_filenames)
331 
332  if len(samples)==0:
333  print "No Samples found... Quitting"
334  return 0
335 
336 # blacklists=guess_blacklists(samples,cmssw_version1,cmssw_version2,options.hlt)
337 
338  # Launch the single comparisons
339  original_dir=os.getcwd()
340 
341  outdir=options.out_dir
342  if len(outdir)==0:
343  print "Creating automatic outdir:",
344  outdir="%sVS%s" %(cmssw_version1,cmssw_version2)
345  print outdir
346  if len(options.input_dir)==0:
347  print "Creating automatic indir:",
348  options.input_dir=outdir
349  print options.input_dir
350 
351  if not os.path.exists(outdir):
352  os.mkdir(outdir)
353  os.chdir(outdir)
354 
355  # adjust the number of threads
356  n_comparisons=len(ref_filenames)
357  if n_comparisons < n_processes:
358  print "Less comparisons than possible processes: reducing n processes to",
359  n_processes=n_comparisons
360  #elif n_processes/n_comparisons == 0:
361  #print "More comparisons than possible processes, can be done in N rounds: reducing n processes to",
362  #original_nprocesses=n_processes
363  #first=True
364  #n_bunches=0
365  #while first or n_processes > original_nprocesses:
366  #n_processes=n_comparisons/2
367  #if n_comparisons%2 !=0:
368  #n_processes+=1
369  #first=False
370 
371  #print n_processes
372  #print n_processes
373 
374  # Test if we treat data
375  skim_name=""
376  if search("20[01]",cmssw_version1)!=None:
377  skim_name=cmssw_version1.split("_")[-1]
378 
379  running_subprocesses=[]
380  process_counter=0
381  #print ref_filenames
382 
383  ## Compare all pairs of root files
384  pool = Pool(n_processes)
385  args_iterable = [list(args) + [options] for args in zip(samples, ref_filenames, test_filenames)]
386  pool.map(call_compare_using_files, args_iterable)
387  # move the pickles on the top, hack
388  os.system("mv */*pkl .")
389 
390  os.chdir("..")
#-------------------------------------------------------------------------------
tuple zip
Definition: archive.py:476
std::vector< T >::const_iterator search(const cond::Time_t &val, const std::vector< T > &container)
Definition: IOVProxy.cc:252
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 ValidationMatrix.do_html (   options,
  hashing_flag,
  standalone 
)

Definition at line 419 of file ValidationMatrix.py.

References do_reports(), and directories2html.make_summary_table().

420 def do_html(options, hashing_flag, standalone):
421 
422  if options.reports:
423  print "Preparing reports for the single files..."
424  do_reports(options.input_dir)
425  # Do the summary page
426  aggregation_rules={}
427  aggregation_rules_twiki={}
428  # check which aggregation rules are to be used
429  if options.hlt:
430  print "Aggregating directories according to HLT rules"
431  aggregation_rules=definitions.aggr_pairs_dict['HLT']
432  aggregation_rules_twiki=definitions.aggr_pairs_twiki_dict['HLT']
433  else:
434  aggregation_rules=definitions.aggr_pairs_dict['reco']
435  aggregation_rules_twiki=definitions.aggr_pairs_twiki_dict['reco']
436  table_html = make_summary_table(options.input_dir,aggregation_rules,aggregation_rules_twiki, hashing_flag, standalone)
437 
438  # create summary html file
439  ofile = open("RelMonSummary.html","w")
440  ofile.write(table_html)
441  ofile.close()
442 
443 #-------------------------------------------------------------------------------
def ValidationMatrix.do_reports (   indir)

Definition at line 391 of file ValidationMatrix.py.

References alcazmumu_cfi.filter.

Referenced by do_html().

392 def do_reports(indir):
393  #print indir
394  os.chdir(indir)
395  pkl_list=filter(lambda x:".pkl" in x, os.listdir("./"))
396  running_subprocesses=[]
397  n_processes=int(options.n_processes)
398  process_counter=0
399  for pklfilename in pkl_list:
400  command = "compare_using_files.py "
401  command+= "-R "
402  if options.do_pngs:
403  command+= " -p "
404  command+= "-P %s " %pklfilename
405  command+= "-o %s " %pklfilename[:-4]
406  print "Executing %s" %command
407  process=call(filter(lambda x: len(x)>0,command.split(" ")))
408  process_counter+=1
409  # add it to the list
410  running_subprocesses.append(process)
411  if process_counter>=n_processes:
412  process_counter=0
413  for p in running_subprocesses:
414  #print "Waiting for %s" %p.name
415  p.wait()
416 
417  os.chdir("..")
418 
#-------------------------------------------------------------------------------
def ValidationMatrix.get_clean_fileanames (   ref_samples,
  test_samples 
)

Definition at line 252 of file ValidationMatrix.py.

References check_root_files(), cmsRelvalreport.exit, and python.multivaluedict.map().

Referenced by do_comparisons_threaded().

253 def get_clean_fileanames(ref_samples,test_samples):
254  # Process the samples starting from the names
255  ref_filenames=map(lambda s:s.strip(),ref_samples.split(","))
256  test_filenames=map(lambda s:s.strip(),test_samples.split(","))
257 
258  if len(ref_filenames)!=len(test_filenames):
259  print "The numebr of reference and test files does not seem to be the same. Please check."
260  exit(2)
261 
262  if not (check_root_files(ref_filenames) and check_root_files(test_filenames)):
263  exit(2)
264  return ref_filenames,test_filenames
265 
266 #-------------------------------------------------------------------------------
def ValidationMatrix.get_filenames_from_pool (   all_samples)

Definition at line 198 of file ValidationMatrix.py.

References get_roofiles_in_dir(), utils.make_files_pairs(), and archive.zip.

Referenced by do_comparisons_threaded().

199 def get_filenames_from_pool(all_samples):
200 
201  # get a list of the files
202  files_list=get_roofiles_in_dir(all_samples)
203 
204  if len(files_list)==0:
205  print "Zero files found in directory %s!" %all_samples
206  return [],[]
207 
208  # Are they an even number?
209  for name in files_list:
210  print "* ",name
211  if len(files_list)%2!=0:
212  print "The numbuer of file is not even... Trying to recover a catastrophe."
213 
214  files_list=make_files_pairs(files_list)
215 
216  # Try to couple them according to their sample
217  ref_filenames=[]
218  test_filenames=[]
219  #files_list.sort(key=name2version)
220  #files_list.sort(key=name2sample)
221  #files_list.sort(key=name2run)
222  for iname in xrange(len(files_list)):
223  filename=files_list[iname]
224  if iname%2==0:
225  ref_filenames.append(filename)
226  else:
227  test_filenames.append(filename)
228 
229  print "The guess would be the following:"
230  for ref,test in zip(ref_filenames,test_filenames):
231  refbasedir=os.path.dirname(ref)
232  testbasedir=os.path.dirname(test)
233  dir_to_print=refbasedir
234  if refbasedir!=testbasedir:
235  dir_to_print="%s and %s" %(refbasedir,testbasedir)
236  print "* Directory: %s " %dir_to_print
237  refname=os.path.basename(ref)
238  testname=os.path.basename(test)
239  print " o %s" %refname
240  print " o %s" %testname
241 
242  #is_ok=ask_ok("Is that ok?")
243  #if not is_ok:
244  #print "Manual input needed then!"
245  #exit(2)
246 
247 
248  return ref_filenames,test_filenames
249 
250 
251 #-------------------------------------------------------------------------------
tuple zip
Definition: archive.py:476
def make_files_pairs
Definition: utils.py:549
def ValidationMatrix.get_roofiles_in_dir (   directory)

Definition at line 189 of file ValidationMatrix.py.

References alcazmumu_cfi.filter, and python.multivaluedict.map().

Referenced by get_filenames_from_pool().

190 def get_roofiles_in_dir(directory):
191  print directory
192  files_list = filter(lambda s: s.endswith(".root"), os.listdir(directory))
193  files_list_path=map(lambda s: os.path.join(directory,s), files_list)
194 
195  return files_list_path
196 
197 #-------------------------------------------------------------------------------
def ValidationMatrix.guess_blacklists (   samples,
  ver1,
  ver2,
  hlt 
)
Build a blacklist for each sample accordind to a set of rules

Definition at line 148 of file ValidationMatrix.py.

References add_to_blacklist(), and cond::persistency.search().

Referenced by call_compare_using_files().

149 def guess_blacklists(samples,ver1,ver2,hlt):
150  """Build a blacklist for each sample accordind to a set of rules
151  """
152  blacklists={}
153  for sample in samples:
154  blacklists[sample]="FED@1,AlcaBeamMonitor@1,HLT@1,AlCaReco@1"
155 
156  # HLT
157  if hlt: #HLT
158  blacklists[sample]+=",AlCaEcalPi0@2"
159  if not search("2010+|2011+|2012+",ver1):
160  print "We are treating MC files for the HLT"
161  for pattern,blist in definitions.hlt_mc_pattern_blist_pairs:
162  blacklists[sample]=add_to_blacklist(blacklists[sample],pattern,sample,blist)
163  else:
164  print "We are treating Data files for the HLT"
165  # at the moment it does not make sense since hlt is ran already
166 
167  else: #RECO
168  #Monte Carlo
169  if not search("2010+|2011+|2012+",ver1):
170  print "We are treating MC files"
171 
172  for pattern,blist in definitions.mc_pattern_blist_pairs:
173  blacklists[sample]=add_to_blacklist(blacklists[sample],pattern,sample,blist)
174 # print "MC RECO"
175  #print blacklists[sample]
176 
177  # Data
178  else:
179  print "We are treating Data files:"
180  blacklists[sample]+=",By__Lumi__Section@-1,AlCaReco@1"
181  for pattern,blist in definitions.data_pattern_blist_pairs:
182  blacklists[sample]=add_to_blacklist(blacklists[sample],pattern,ver1,blist)
183 # print "DATA RECO: %s %s %s -->%s" %( ver1, pattern, blist, blacklists[sample])
184 
185 
186  return blacklists
187 
188 #-------------------------------------------------------------------------------
std::vector< T >::const_iterator search(const cond::Time_t &val, const std::vector< T > &container)
Definition: IOVProxy.cc:252
def ValidationMatrix.guess_params (   ref_filenames,
  test_filenames 
)

Definition at line 67 of file ValidationMatrix.py.

References list(), python.multivaluedict.map(), name2sample(), name2version(), cond::persistency.search(), and archive.zip.

Referenced by do_comparisons_threaded().

67 
68 def guess_params(ref_filenames,test_filenames):
69 
70  if len(ref_filenames)*len(test_filenames)==0:
71  print "Empty reference and test filenames lists!"
72  return [],"",""
73 
74  samples=[]
75  ref_versions=[]
76  test_versions=[]
77 
78  for ref, test in zip(map(os.path.basename,ref_filenames),map(os.path.basename,test_filenames)):
79 
80  ref_sample=name2sample(ref)
81  ref_version=name2version(ref)
82  test_sample=name2sample(test)
83  test_version=name2version(test)
84 
85  print " ## sample 1: %s vs sample 2: %s"%(ref_sample, test_sample)
86 
87  if ref_sample!=test_sample:
88  print "Files %s and %s do not seem to be relative to the same sample." %(ref, test)
89  # exit(2)
90 
91  # Slightly modify for data
92  if search("20[01]",ref_version)!=None:
93  ref_sample+=ref_version.split("_")[-1]
94  samples.append(ref_sample)
95 
96  # append the versions
97  ref_versions.append(ref_version)
98  test_versions.append(test_version)
99 
100  # Check if ref and test versions are always the same.
101  ref_versions=list(set(ref_versions))
102  test_versions=list(set(test_versions))
103 
104  #for versions in ref_versions,test_versions:
105  #if len(versions)!=1:
106  #print "More than one kind of CMSSW version selected (%s)" %versions
107  #exit(2)
108 
109  cmssw_version1=ref_versions[0]
110  cmssw_version2=test_versions[0]
111 
112  return samples,cmssw_version1,cmssw_version2
113 
114 
115 #-------------------------------------------------------------------------------
tuple zip
Definition: archive.py:476
std::vector< T >::const_iterator search(const cond::Time_t &val, const std::vector< T > &container)
Definition: IOVProxy.cc:252
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 ValidationMatrix.name2globaltag (   filename)

Definition at line 61 of file ValidationMatrix.py.

References split.

Referenced by call_compare_using_files().

61 
62 def name2globaltag(filename):
63  namebase = os.path.basename(filename)
64  return namebase.split("__")[2].split("-")[1] #returns GT from file basename
65 
66 #-------------------------------------------------------------------------------
double split
Definition: MVATrainer.cc:139
def ValidationMatrix.name2run (   filename)

Definition at line 49 of file ValidationMatrix.py.

References split.

Referenced by name2runskim().

49 
50 def name2run(filename):
51  namebase=os.path.basename(filename)
52  return namebase.split("__")[0].split("_")[2]
double split
Definition: MVATrainer.cc:139
def ValidationMatrix.name2runskim (   filename)

Definition at line 53 of file ValidationMatrix.py.

References name2run(), name2version(), and split.

53 
54 def name2runskim(filename):
55  run=name2run(filename)
56  skim=name2version(filename).split("_")[-1]
57  # remove skim version
58  if "-v" in skim:
59  skim = skim[:skim.rfind('-v')]
60  return "%s_%s"%(run,skim)
double split
Definition: MVATrainer.cc:139
def ValidationMatrix.name2sample (   filename)

Definition at line 41 of file ValidationMatrix.py.

Referenced by guess_params().

41 
42 def name2sample(filename):
43  namebase=os.path.basename(filename)
44  return namebase.split("__")[1]
def ValidationMatrix.name2version (   filename)

Definition at line 45 of file ValidationMatrix.py.

Referenced by call_compare_using_files(), guess_params(), and name2runskim().

45 
46 def name2version(filename):
47  namebase=os.path.basename(filename)
48  return namebase.split("__")[2]

Variable Documentation

string ValidationMatrix.action = "store"

—HASHING—##

– USE CSS files in web access, for stand-alone usage –##

–Blacklist File –##

Definition at line 464 of file ValidationMatrix.py.

string ValidationMatrix.all_samples = ""

Definition at line 449 of file ValidationMatrix.py.

ValidationMatrix.argv = sys.argv

Definition at line 23 of file ValidationMatrix.py.

ValidationMatrix.default = ref_samples,

Definition at line 466 of file ValidationMatrix.py.

string ValidationMatrix.dest = "ref_samples"

Definition at line 465 of file ValidationMatrix.py.

string ValidationMatrix.help = "The samples that act as reference (comma separated list)"

Definition at line 467 of file ValidationMatrix.py.

ValidationMatrix.hlt = False

Definition at line 457 of file ValidationMatrix.py.

string ValidationMatrix.in_dir = ""

Definition at line 452 of file ValidationMatrix.py.

int ValidationMatrix.n_processes = 1

Definition at line 450 of file ValidationMatrix.py.

int ValidationMatrix.n_threads = 1

Definition at line 453 of file ValidationMatrix.py.

string ValidationMatrix.out_dir = ""

Definition at line 451 of file ValidationMatrix.py.

Referenced by HtrXmlPattern.HtrXmlPattern().

tuple ValidationMatrix.parser = OptionParser(usage="usage: %prog [options]")

Definition at line 461 of file ValidationMatrix.py.

string ValidationMatrix.ref_samples = ""

Definition at line 447 of file ValidationMatrix.py.

int ValidationMatrix.run = -1

Definition at line 454 of file ValidationMatrix.py.

string ValidationMatrix.stat_test = "Chi2"

Definition at line 455 of file ValidationMatrix.py.

string ValidationMatrix.test_samples = ""

Definition at line 448 of file ValidationMatrix.py.

float ValidationMatrix.test_threshold = 0.00001

Definition at line 456 of file ValidationMatrix.py.