def readPrimaryDataset(prdsName): priDSVector = dbsclient.PriDSVector() client.readPrimaryDataset(prdsName, priDSVector) print "no of Primary Datasets ",priDSVector.size() for i in range( priDSVector.size() ): print "************************************************************************************************" primaryDsId = dbsclient.intp_value(priDSVector[i].t_primary_dataset_id.getValue()) print "primary_dataset_id = ",primaryDsId print "primary_dataset_name = ",dbsclient.stringp_value(priDSVector[i].t_primary_dataset_name.getValue()) print "primary_dataset_description = ",dbsclient.intp_value(priDSVector[i].t_primary_dataset_description.getValue()) readProcessingPath(primaryDsId)
def readEvCollFiles(evCollId): evCollFileVector = dbsclient.EVCollFileVector() client.readEvCollFiles(evCollId, evCollFileVector) print "no of Files in Event Collection",evCollFileVector.size() for i in range( evCollFileVector.size() ): print "************************************************************************************************" print "file_logical_name = ",dbsclient.stringp_value(evCollFileVector[i].t_file_logical_name.getValue()) print "file_guid = ",dbsclient.stringp_value(evCollFileVector[i].t_file_guid.getValue()) print "file_inblock = ",dbsclient.intp_value(evCollFileVector[i].t_file_inblock.getValue())
def readProcessingPath(primaryDsId): proPathVector = dbsclient.ProPathVector() client.readProcessingPath(primaryDsId, proPathVector) print "no of Processing Path ",proPathVector.size() for i in range( proPathVector.size() ): print "************************************************************************************************" proDsId = dbsclient.intp_value(proPathVector[i].t_processed_dataset_id.getValue()) print "processed_dataset_id = ",proDsId print "data tiername = ",dbsclient.stringp_value(proPathVector[i].t_data_tier_name.getValue()) readEvColls(proDsId)
def readEvColls(proDsId): evCollVector = dbsclient.EVCollVector() client.readEvColls(proDsId, evCollVector) print "no of Event Collections ",evCollVector.size() for i in range( evCollVector.size() ): evCollId = dbsclient.intp_value(evCollVector[i].t_event_collection_id.getValue()) print "************************************************************************************************" print "status = ",dbsclient.intp_value(evCollVector[i].t_info_evcoll_status.getValue()) print "validation_status = ",dbsclient.intp_value(evCollVector[i].t_info_evcoll_validation_status.getValue()) print "processed_dataset = ",dbsclient.intp_value(evCollVector[i].t_event_collection_processed_dataset.getValue()) print "events = ",dbsclient.intp_value(evCollVector[i].t_info_evcoll_events.getValue()) print "collection_index = ",dbsclient.intp_value(evCollVector[i].t_event_collection_collection_index.getValue()) print "id = ",evCollId print "name = ",dbsclient.stringp_value(evCollVector[i].t_info_evcoll_name.getValue()) readEvCollFiles(evCollId)
def readEvCollFiles(self, evCollId):
filesByBlock = {}
evCollFileVector = dbsclient.EVCollFileVector()
evCollInfo = dbsclient.Fileview_ClientAPIData()
evCollInfo.t_evcoll_file_evcoll = dbsclient.AINT(evCollId)
self.client.readEvCollFiles(evCollInfo, evCollFileVector)
print "no of Files in Event Collection",evCollFileVector.size()
for i in range( evCollFileVector.size() ):
print "************************************************************************************************"
lfn = dbsclient.stringp_value(evCollFileVector[i].t_file_logical_name.getValue())
guid = dbsclient.stringp_value(evCollFileVector[i].t_file_guid.getValue())
inBlock = dbsclient.intp_value(evCollFileVector[i].t_file_inblock.getValue())
print "file_logical_name = ", lfn
print "file_guid = ", guid
print "file_inblock = ", inBlock
if not filesByBlock.has_key(inBlock):
filesByBlock[inBlock] = [(lfn, guid)]
else :
filesByBlock[inBlock].append((lfn, guid))
return filesByBlock
def readEvColls(self, proDsId):
evCollInfoList = []
evCollVector = dbsclient.EVCollVector()
evCollInfo = dbsclient.Evcollview_ClientAPIData()
evCollInfo.t_event_collection_processed_dataset = dbsclient.AINT(proDsId)
self.client.readEvColls(evCollInfo, evCollVector)
print "no of Event Collections ",evCollVector.size()
for i in range( evCollVector.size() ):
evCollId = dbsclient.intp_value(evCollVector[i].t_event_collection_id.getValue())
events = dbsclient.intp_value(evCollVector[i].t_info_evcoll_events.getValue())
name = dbsclient.stringp_value(evCollVector[i].t_info_evcoll_name.getValue())
evcollInfo = (evCollId, events, name)
evCollInfoList.append(evcollInfo)
print "************************************************************************************************"
print "status = ",dbsclient.intp_value(evCollVector[i].t_info_evcoll_status.getValue())
print "validation_status = ",dbsclient.intp_value(evCollVector[i].t_info_evcoll_validation_status.getValue())
print "processed_dataset = ",dbsclient.intp_value(evCollVector[i].t_event_collection_processed_dataset.getValue())
print "events = ", events
print "collection_index = ",dbsclient.intp_value(evCollVector[i].t_event_collection_collection_index.getValue())
print "id = ",evCollId
print "name = ", name
return evCollInfoList
def getDatasetProvenance(self, pathName, parentDataTiers):
""" API Call that returns list of DBS
Datasets (Processed/Analysis) corresponding
to all parent datatiers requested by user
"""
raise dbsApi.DbsApiException(args="getDatasetProvenance() API Call Not Implemented")
# get the names of primary dataset, data tier and processed dataset
tokens = pathName.split('/')
primaryDSName = tokens[1]
dataTier = tokens[2]
processedDSName = tokens[3]
#Get the list existing Parent DataTiers
provInfo = dbsclient.Datasetprovenenceevchild_ClientAPIData()
provInfo.t_primary_dataset_name = dbsclient.ASTR(primaryDSName)
provInfo.t_data_tier_name = dbsclient.ASTR(dataTier)
provInfo.t_processed_dataset_name = dbsclient.ASTR(processedDSName)
childIds=[]
provInfoRet = self.readProvInfoChild(provInfo)
for i in range( provInfoRet.size() ) :
childId = dbsclient.intp_value(provInfoRet[i].t_evcoll_parentage_child.getValue())
if childId not in childIds:
childIds.append(childId)
#print "CHILD IDs ", childIds
if len(childIds) == 0 :
errorMessage = "No Parents found for this Dataset " + pathName
raise dbsApi.DbsApiException(args=errorMessage)
provInfo = dbsclient.Datasetprovenenceevparent_ClientAPIData()
#provInfo.t_evcoll_parentage_child = dbsclient.AINT(224)
provInfo.t_evcoll_parentage_child = dbsclient.AINT(childIds[0])
provInfoRet = self.readProvInfoParent(provInfo)
print "After self.readProvInfoParent provInfoRet is ",provInfoRet
dt_procDS_map = {}
for i in range( provInfoRet.size() ) :
print "************************************************************************************************"
currProvInfo = provInfoRet[i]
processed_dataset_name = dbsclient.stringp_value(currProvInfo.t_processed_dataset_name.getValue())
print " t_processed_dataset_name: " ,processed_dataset_name
#print " t_processing_path_data_tier: ", dbsclient.intp_value(currProvInfo.t_processing_path_data_tier.getValue())
#print " t_processing_path_id: ", dbsclient.intp_value(currProvInfo.t_processing_path_id.getValue())
parentage_type_name = dbsclient.stringp_value(currProvInfo.t_parentage_type_name.getValue())
print " t_parentage_type_name: " ,parentage_type_name
primaryDsId = dbsclient.intp_value(currProvInfo.t_primary_dataset_id.getValue())
print " t_primary_dataset_id: " , primaryDsId
#print " t_processing_path_full_path: " ,dbsclient.stringp_value(currProvInfo.t_processing_path_full_path.getValue())
#print " t_primary_dataset_name: ",dbsclient.stringp_value(currProvInfo.t_primary_dataset_name.getValue())
#print " t_evcoll_parentage_parent: ", dbsclient.intp_value(currProvInfo.t_evcoll_parentage_parent.getValue())
procDSId = dbsclient.intp_value(currProvInfo.t_processed_dataset_id.getValue())
print " t_processed_dataset_id: ", procDSId
dTier=dbsclient.stringp_value(currProvInfo.t_data_tier_name.getValue())
print " t_data_tier.name: ", dTier
#print " t_evcoll_parentage_child: ", dbsclient.intp_value(currProvInfo.t_evcoll_parentage_child.getValue())
if not dt_procDS_map.has_key(dTier):
dt_procDS_map[dTier]=processed_dataset_name, parentage_type_name
#if not dt_procDS_map.has_key(data_tier_name):
# dt_procDS_map[data_tier_name]=processed_dataset_name
parentPathList = []
dsList = DbsDatasetList()
print "dt_procDS_map: ", dt_procDS_map
if len(dt_procDS_map) < 1:
raise dbsApi.DbsApiException(args="No parent path found")
for aAskedTier in parentDataTiers:
l_processed_dataset_name, l_parentage_type_name = dt_procDS_map[aAskedTier]
#parentPath = "/"+primaryDSName+"/"+aAskedTier+"/"+dt_procDS_map[aAskedTier]
parentPath = "/"+primaryDSName+"/"+aAskedTier+"/"+l_processed_dataset_name
dsList.append(DbsDataset(primaryDSName, parentPath, aAskedTier, l_parentage_type_name))
parentPathList.append(parentPath)
print parentPath
#return parentPathList
return dsList
