| 39 |
|
# #################################### |
| 40 |
|
# class to find and extact info from published data |
| 41 |
|
class DataDiscovery: |
| 42 |
< |
def __init__(self, owner, dataset, dataTiers, cfg_params): |
| 42 |
> |
def __init__(self, datasetPath, dataTiers, cfg_params): |
| 43 |
|
|
| 44 |
|
# Attributes |
| 45 |
< |
self.owner = owner |
| 46 |
< |
self.dataset = dataset |
| 45 |
> |
self.datasetPath = datasetPath |
| 46 |
|
self.dataTiers = dataTiers |
| 47 |
|
self.cfg_params = cfg_params |
| 48 |
|
|
| 49 |
< |
self.dbspaths= [] # DBS output: list of dbspaths for all data |
| 50 |
< |
self.allblocks = [] # DBS output: list of map fileblocks-totevts for all dataset-owners |
| 51 |
< |
self.blocksinfo = {} # DBS output: map fileblocks-totevts for the primary block, used internally to this class |
| 49 |
> |
self.eventsPerBlock = {} # DBS output: map fileblocks-events for collection |
| 50 |
> |
self.eventsPerFile = {} # DBS output: map files-events |
| 51 |
> |
self.blocksinfo = {} # DBS output: map fileblocks-files |
| 52 |
|
#DBS output: max events computed by method getMaxEvents |
| 53 |
|
|
| 54 |
|
# #################################### |
| 57 |
|
Contact DBS |
| 58 |
|
""" |
| 59 |
|
|
| 60 |
< |
## add the PU among the required data tiers if the Digi are requested |
| 62 |
< |
if (self.dataTiers.count('Digi')>0) & (self.dataTiers.count('PU')<=0) : |
| 63 |
< |
self.dataTiers.append('PU') |
| 64 |
< |
|
| 65 |
< |
## get info about the requested dataset |
| 66 |
< |
dbs=DBSInfo() |
| 60 |
> |
## get DBS URL |
| 61 |
|
try: |
| 62 |
< |
self.datasets = dbs.getMatchingDatasets(self.owner, self.dataset) |
| 63 |
< |
except DBSError, ex: |
| 64 |
< |
raise DataDiscoveryError(ex.getErrorMessage()) |
| 71 |
< |
if len(self.datasets) == 0: |
| 72 |
< |
raise DataDiscoveryError("Owner=%s, Dataset=%s unknown to DBS" % (self.owner, self.dataset)) |
| 73 |
< |
if len(self.datasets) > 1: |
| 74 |
< |
raise DataDiscoveryError("Owner=%s, Dataset=%s is ambiguous" % (self.owner, self.dataset)) |
| 75 |
< |
try: |
| 76 |
< |
self.dbsdataset = self.datasets[0].get('datasetPathName') |
| 77 |
< |
self.blocksinfo = dbs.getDatasetContents(self.dbsdataset) |
| 78 |
< |
self.allblocks.append (self.blocksinfo.keys ()) # add also the current fileblocksinfo |
| 79 |
< |
self.dbspaths.append(self.dbsdataset) |
| 80 |
< |
except DBSError, ex: |
| 81 |
< |
raise DataDiscoveryError(ex.getErrorMessage()) |
| 82 |
< |
|
| 83 |
< |
if len(self.blocksinfo)<=0: |
| 84 |
< |
msg="\nERROR Data for %s do not exist in DBS! \n Check the dataset/owner variables in crab.cfg !"%self.dbsdataset |
| 85 |
< |
raise NotExistingDatasetError(msg) |
| 62 |
> |
dbs_url=self.cfg_params['CMSSW.dbs_url'] |
| 63 |
> |
except KeyError: |
| 64 |
> |
dbs_url="http://cmsdoc.cern.ch/cms/test/aprom/DBS/CGIServer/prodquery" |
| 65 |
|
|
| 66 |
< |
|
| 67 |
< |
## get info about the parents |
| 66 |
> |
## get info about the requested dataset |
| 67 |
> |
try: |
| 68 |
> |
dbs_instance=self.cfg_params['CMSSW.dbs_instance'] |
| 69 |
> |
except KeyError: |
| 70 |
> |
dbs_instance="MCGlobal/Writer" |
| 71 |
> |
|
| 72 |
> |
dbs = DBSInfo(dbs_url, dbs_instance) |
| 73 |
|
try: |
| 74 |
< |
parents=dbs.getDatasetProvenance(self.dbsdataset, self.dataTiers) |
| 75 |
< |
except DBSInvalidDataTierError, ex: |
| 92 |
< |
msg=ex.getErrorMessage()+' \n Check the data_tier variable in crab.cfg !\n' |
| 74 |
> |
self.datasets = dbs.getMatchingDatasets(self.datasetPath) |
| 75 |
> |
except dbsCgiApi.DbsCgiExecutionError, msg: |
| 76 |
|
raise DataDiscoveryError(msg) |
| 94 |
– |
except DBSError, ex: |
| 95 |
– |
raise DataDiscoveryError(ex.getErrorMessage()) |
| 77 |
|
|
| 78 |
< |
## check that the user asks for parent Data Tier really existing in the DBS provenance |
| 79 |
< |
self.checkParentDataTier(parents, self.dataTiers) |
| 78 |
> |
if len(self.datasets) == 0: |
| 79 |
> |
raise DataDiscoveryError("DatasetPath=%s unknown to DBS" %self.datasetPath) |
| 80 |
> |
if len(self.datasets) > 1: |
| 81 |
> |
raise DataDiscoveryError("DatasetPath=%s is ambiguous" %self.datasetPath) |
| 82 |
|
|
| 100 |
– |
## for each parent get the corresponding fileblocks |
| 83 |
|
try: |
| 84 |
< |
for p in parents: |
| 85 |
< |
## fill a list of dbspaths |
| 86 |
< |
parentPath = p.get('parent').get('datasetPathName') |
| 87 |
< |
self.dbspaths.append (parentPath) |
| 88 |
< |
parentBlocks = dbs.getDatasetContents (parentPath) |
| 107 |
< |
self.allblocks.append (parentBlocks.keys ()) # add parent fileblocksinfo |
| 84 |
> |
self.dbsdataset = self.datasets[0].get('datasetPathName') |
| 85 |
> |
|
| 86 |
> |
self.eventsPerBlock = dbs.getEventsPerBlock(self.dbsdataset) |
| 87 |
> |
self.blocksinfo = dbs.getDatasetFileBlocks(self.dbsdataset) |
| 88 |
> |
self.eventsPerFile = dbs.getEventsPerFile(self.dbsdataset) |
| 89 |
|
except DBSError, ex: |
| 90 |
|
raise DataDiscoveryError(ex.getErrorMessage()) |
| 91 |
< |
|
| 92 |
< |
# ################################################# |
| 93 |
< |
def checkParentDataTier(self, parents, dataTiers): |
| 94 |
< |
""" |
| 95 |
< |
check that the data tiers requested by the user really exists in the provenance of the given dataset |
| 115 |
< |
""" |
| 116 |
< |
startType = string.split(self.dbsdataset,'/')[2] |
| 117 |
< |
# for example 'type' is PU and 'dataTier' is Hit |
| 118 |
< |
parentTypes = map(lambda p: p.get('type'), parents) |
| 119 |
< |
for tier in dataTiers: |
| 120 |
< |
if parentTypes.count(tier) <= 0 and tier != startType: |
| 121 |
< |
msg="\nERROR Data %s not published in DBS with asked data tiers : the data tier not found is %s !\n Check the data_tier variable in crab.cfg !"%(self.dbsdataset,tier) |
| 122 |
< |
raise NoDataTierinProvenanceError(msg) |
| 91 |
> |
|
| 92 |
> |
if len(self.eventsPerBlock) <= 0: |
| 93 |
> |
raise NotExistingDatasetError (("\nNo data for %s in DBS\nPlease check" |
| 94 |
> |
+ " dataset path variables in crab.cfg") |
| 95 |
> |
% self.dbsdataset) |
| 96 |
|
|
| 97 |
|
|
| 98 |
|
# ################################################# |
| 99 |
|
def getMaxEvents(self): |
| 100 |
|
""" |
| 101 |
< |
max events of the primary dataset-owner |
| 101 |
> |
max events |
| 102 |
|
""" |
| 103 |
< |
## loop over the fileblocks of the primary dataset-owner |
| 103 |
> |
## loop over the event collections |
| 104 |
|
nevts=0 |
| 105 |
< |
for blockevts in self.blocksinfo.values(): |
| 106 |
< |
nevts=nevts+blockevts |
| 105 |
> |
for evc_evts in self.eventsPerBlock.values(): |
| 106 |
> |
nevts=nevts+evc_evts |
| 107 |
|
|
| 108 |
|
return nevts |
| 109 |
|
|
| 110 |
|
# ################################################# |
| 111 |
< |
def getDBSPaths(self): |
| 111 |
> |
def getEventsPerBlock(self): |
| 112 |
|
""" |
| 113 |
< |
list the DBSpaths for all required data |
| 113 |
> |
list the event collections structure by fileblock |
| 114 |
|
""" |
| 115 |
< |
return self.dbspaths |
| 115 |
> |
return self.eventsPerBlock |
| 116 |
|
|
| 117 |
|
# ################################################# |
| 118 |
< |
def getEVC(self): |
| 118 |
> |
def getEventsPerFile(self): |
| 119 |
|
""" |
| 120 |
< |
list the event collections structure by fileblock |
| 120 |
> |
list the event collections structure by file |
| 121 |
|
""" |
| 122 |
< |
print "To be used by a more complex job splitting... TODO later... " |
| 150 |
< |
print "it requires changes in what's returned by DBSInfo.getDatasetContents and then fetchDBSInfo" |
| 122 |
> |
return self.eventsPerFile |
| 123 |
|
|
| 124 |
|
# ################################################# |
| 125 |
< |
def getFileBlocks(self): |
| 125 |
> |
def getFiles(self): |
| 126 |
|
""" |
| 127 |
< |
fileblocks for all required dataset-owners |
| 127 |
> |
return files grouped by fileblock |
| 128 |
|
""" |
| 129 |
< |
return self.allblocks |
| 129 |
> |
return self.blocksinfo |
| 130 |
|
|
| 131 |
|
######################################################################## |
