| 1 |
afanfani |
1.1 |
#!/usr/bin/env python
|
| 2 |
|
|
import sys, os, string, re
|
| 3 |
|
|
import xml.sax
|
| 4 |
|
|
import urllib
|
| 5 |
|
|
import DBSXMLHandler
|
| 6 |
|
|
|
| 7 |
|
|
class DBSError:
|
| 8 |
|
|
def __init__(self, owner, dataset):
|
| 9 |
|
|
print '\nERROR accessing DBS for Owner/Dataset: '+owner+'/'+dataset+'\n'
|
| 10 |
|
|
pass
|
| 11 |
|
|
|
| 12 |
|
|
class DBSInfoError:
|
| 13 |
|
|
def __init__(self, url):
|
| 14 |
|
|
print '\nERROR accessing DBS url : '+url+'\n'
|
| 15 |
|
|
pass
|
| 16 |
|
|
|
| 17 |
|
|
##################################################################################
|
| 18 |
|
|
# Class to extract info from DBS
|
| 19 |
|
|
###############################################################################
|
| 20 |
|
|
|
| 21 |
|
|
class DBSInfo:
|
| 22 |
|
|
def __init__(self, owner, dataset, dataTiers):
|
| 23 |
|
|
self.owner = owner
|
| 24 |
|
|
self.dataset = dataset
|
| 25 |
|
|
self.dataTiers = dataTiers
|
| 26 |
|
|
## DBSXMLDump
|
| 27 |
|
|
self.DBSclient_ = '/phedex/SL/PHEDEX_head/Utilities/DBSXMLDump -from DBS '
|
| 28 |
|
|
self.DBSParam_ = '~/.globus/DBS_DBParam:Production/Admin'
|
| 29 |
|
|
self.DBSDumpbase_ = self.DBSclient_+" -from DBS -db "+self.DBSParam_
|
| 30 |
|
|
## HTTP/CGI
|
| 31 |
|
|
self.DBSURL_='http://cern.ch/cms-dbs/cgi-bin/dbsxml?api='
|
| 32 |
|
|
|
| 33 |
|
|
# ####################################
|
| 34 |
|
|
def getDatasetProvenance(self):
|
| 35 |
|
|
"""
|
| 36 |
|
|
query DBS to get provenance
|
| 37 |
|
|
"""
|
| 38 |
|
|
## DBSXMLDump
|
| 39 |
|
|
xmlfile= 'parent.xml'
|
| 40 |
|
|
ListdataTiers=string.join(self.dataTiers,',')
|
| 41 |
|
|
#print ListdataTiers
|
| 42 |
|
|
print "DBSInfo: ---> getDatasetProvenance : "+self.owner+"/"+self.dataset+" with datatier "+ListdataTiers
|
| 43 |
|
|
cmd = self.DBSDumpbase_+" -to "+xmlfile+" -datatier "+ListdataTiers+" -getDatasetProvenance \'"+self.owner+"/"+self.dataset+"\'"
|
| 44 |
|
|
#print "DBSInfo: executing "+cmd
|
| 45 |
|
|
os.system(cmd)
|
| 46 |
|
|
|
| 47 |
|
|
### parse the XML
|
| 48 |
|
|
handler = self.XMLparsing(xmlfile)
|
| 49 |
|
|
os.system('rm '+xmlfile)
|
| 50 |
|
|
|
| 51 |
|
|
print "DBSInfo: parents are %s"%handler.parentsList
|
| 52 |
|
|
|
| 53 |
|
|
for aparent in handler.parent.keys():
|
| 54 |
|
|
print "DBSInfo: parent is "+aparent+" type: "+handler.parent[aparent]
|
| 55 |
|
|
|
| 56 |
|
|
## return a map with parent path and type : should I return just a list with parent paths?
|
| 57 |
|
|
return handler.parent
|
| 58 |
|
|
|
| 59 |
|
|
|
| 60 |
|
|
# ####################################
|
| 61 |
|
|
def getDatasetContents(self):
|
| 62 |
|
|
"""
|
| 63 |
|
|
query DBS to get event collections
|
| 64 |
|
|
"""
|
| 65 |
|
|
print "DBSInfo: ---> getDatasetContents : "+self.dataset+"/"+self.owner
|
| 66 |
|
|
try:
|
| 67 |
|
|
url=self.DBSURL_+'getDatasetContents&path='+self.dataset+'/datatier/'+self.owner
|
| 68 |
|
|
f = urllib.urlopen(url)
|
| 69 |
|
|
except:
|
| 70 |
|
|
raise DBSInfoError(url)
|
| 71 |
|
|
|
| 72 |
|
|
data = f.read()
|
| 73 |
|
|
xmlfile = "evc.xml"
|
| 74 |
|
|
file = open(xmlfile, 'w')
|
| 75 |
|
|
file.write(data)
|
| 76 |
|
|
file.close()
|
| 77 |
|
|
|
| 78 |
|
|
### parse the XML
|
| 79 |
|
|
handler = self.XMLparsing(xmlfile)
|
| 80 |
|
|
os.system('rm '+xmlfile)
|
| 81 |
|
|
|
| 82 |
|
|
## get the fileblock and event collections
|
| 83 |
|
|
nevtsbyblock= {}
|
| 84 |
|
|
print "DBSInfo: fileblocks are: %s"%handler.fileblocksList
|
| 85 |
|
|
|
| 86 |
|
|
for block in handler.fileblocksList:
|
| 87 |
|
|
print "DBSInfo: --- block: "+block
|
| 88 |
|
|
## get the event collections for each block
|
| 89 |
|
|
evcnames=handler.evcbyfileblock[block].keys()
|
| 90 |
|
|
evcnames.sort()
|
| 91 |
|
|
nevts=0
|
| 92 |
|
|
for evcname in evcnames:
|
| 93 |
|
|
print "DBSInfo: evc: "+evcname+" nevts: "+handler.evcbyfileblock[block][evcname]
|
| 94 |
|
|
nevts=nevts+int(handler.evcoll[evcname])
|
| 95 |
|
|
print "DBSInfo: total nevts %i in block %s "%(nevts,block)
|
| 96 |
|
|
nevtsbyblock[block]=nevts
|
| 97 |
|
|
|
| 98 |
|
|
print "DBSInfo: total number of events %i"%handler.totnevts
|
| 99 |
|
|
# returning a map of fileblock-nevts will be enough for now
|
| 100 |
|
|
# TODO: in future the EvC collections grouped by fileblock should be returned
|
| 101 |
|
|
|
| 102 |
|
|
return nevtsbyblock
|
| 103 |
|
|
|
| 104 |
|
|
|
| 105 |
|
|
# ####################################
|
| 106 |
|
|
def XMLparsing(self, xmlfile):
|
| 107 |
|
|
"""
|
| 108 |
|
|
parse XML
|
| 109 |
|
|
"""
|
| 110 |
|
|
#print "\n DBSInfo: Parsing XML file "+xmlfile+"\n"
|
| 111 |
|
|
parser = xml.sax.make_parser()
|
| 112 |
|
|
handler = DBSXMLHandler.Handler()
|
| 113 |
|
|
parser.setContentHandler(handler)
|
| 114 |
|
|
parser.parse(xmlfile)
|
| 115 |
|
|
return handler
|
| 116 |
|
|
|
