| 1 |
afanfani |
1.1 |
#!/usr/bin/env python
|
| 2 |
|
|
import sys, os, string, re
|
| 3 |
|
|
import xml.sax
|
| 4 |
|
|
import urllib
|
| 5 |
afanfani |
1.2 |
|
| 6 |
|
|
import sys
|
| 7 |
|
|
sys.path.append('./DBSAPI')
|
| 8 |
|
|
import dbsCgiApi
|
| 9 |
afanfani |
1.1 |
|
| 10 |
|
|
class DBSError:
|
| 11 |
|
|
def __init__(self, owner, dataset):
|
| 12 |
|
|
print '\nERROR accessing DBS for Owner/Dataset: '+owner+'/'+dataset+'\n'
|
| 13 |
|
|
pass
|
| 14 |
|
|
|
| 15 |
|
|
class DBSInfoError:
|
| 16 |
|
|
def __init__(self, url):
|
| 17 |
|
|
print '\nERROR accessing DBS url : '+url+'\n'
|
| 18 |
|
|
pass
|
| 19 |
|
|
|
| 20 |
|
|
##################################################################################
|
| 21 |
|
|
# Class to extract info from DBS
|
| 22 |
|
|
###############################################################################
|
| 23 |
|
|
|
| 24 |
|
|
class DBSInfo:
|
| 25 |
|
|
def __init__(self, owner, dataset, dataTiers):
|
| 26 |
|
|
self.owner = owner
|
| 27 |
|
|
self.dataset = dataset
|
| 28 |
|
|
self.dataTiers = dataTiers
|
| 29 |
afanfani |
1.2 |
self.dbspath=dataset+'/datatier/'+owner
|
| 30 |
|
|
|
| 31 |
|
|
self.api = dbsCgiApi.DbsCgiApi(cgiUrl="http://cern.ch/cms-dbs/cgi-bin")
|
| 32 |
afanfani |
1.1 |
|
| 33 |
|
|
# ####################################
|
| 34 |
|
|
def getDatasetProvenance(self):
|
| 35 |
|
|
"""
|
| 36 |
|
|
query DBS to get provenance
|
| 37 |
|
|
"""
|
| 38 |
afanfani |
1.2 |
datasetParentList = self.api.getDatasetProvenance(self.dbspath,self.dataTiers)
|
| 39 |
|
|
|
| 40 |
|
|
parent = {}
|
| 41 |
|
|
for aparent in datasetParentList:
|
| 42 |
|
|
print "DBSInfo: parent path is "+aparent.getDatasetPath()+" datatier is: "+aparent.getDataTier()
|
| 43 |
|
|
parent[aparent.getDatasetPath()]=aparent.getDataTier()
|
| 44 |
afanfani |
1.1 |
|
| 45 |
afanfani |
1.2 |
return parent
|
| 46 |
afanfani |
1.1 |
|
| 47 |
|
|
# ####################################
|
| 48 |
|
|
def getDatasetContents(self):
|
| 49 |
|
|
"""
|
| 50 |
|
|
query DBS to get event collections
|
| 51 |
|
|
"""
|
| 52 |
afanfani |
1.2 |
|
| 53 |
|
|
fileBlockList = self.api.getDatasetContents(self.dbspath)
|
| 54 |
afanfani |
1.1 |
|
| 55 |
|
|
## get the fileblock and event collections
|
| 56 |
|
|
nevtsbyblock= {}
|
| 57 |
afanfani |
1.2 |
for fileBlock in fileBlockList:
|
| 58 |
|
|
## get the event collections for each block
|
| 59 |
|
|
#print fileBlock.getBlockName()
|
| 60 |
|
|
#print fileBlock.getBlockId()
|
| 61 |
|
|
eventCollectionList = fileBlock.getEventCollectionList()
|
| 62 |
|
|
nevts=0
|
| 63 |
|
|
for eventCollection in eventCollectionList:
|
| 64 |
|
|
#print "DBSInfo: evc: "+eventCollection.getCollectionName()+" nevts: %i"%eventCollection.getNumberOfEvents()
|
| 65 |
|
|
nevts=nevts+eventCollection.getNumberOfEvents()
|
| 66 |
|
|
print "DBSInfo: total nevts %i in block %s "%(nevts,fileBlock.getBlockName())
|
| 67 |
|
|
nevtsbyblock[fileBlock.getBlockName()]=nevts
|
| 68 |
afanfani |
1.1 |
|
| 69 |
|
|
# returning a map of fileblock-nevts will be enough for now
|
| 70 |
|
|
# TODO: in future the EvC collections grouped by fileblock should be returned
|
| 71 |
afanfani |
1.2 |
|
| 72 |
afanfani |
1.1 |
return nevtsbyblock
|
| 73 |
|
|
|
