| 2 |
|
import sys, os, string, re |
| 3 |
|
import xml.sax |
| 4 |
|
import urllib |
| 5 |
< |
import DBSXMLHandler |
| 5 |
> |
|
| 6 |
> |
import sys |
| 7 |
> |
sys.path.append('./DBSAPI') |
| 8 |
> |
import dbsCgiApi |
| 9 |
|
|
| 10 |
|
class DBSError: |
| 11 |
|
def __init__(self, owner, dataset): |
| 26 |
|
self.owner = owner |
| 27 |
|
self.dataset = dataset |
| 28 |
|
self.dataTiers = dataTiers |
| 29 |
< |
## DBSXMLDump |
| 30 |
< |
self.DBSclient_ = '/phedex/SL/PHEDEX_head/Utilities/DBSXMLDump -from DBS ' |
| 31 |
< |
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=' |
| 29 |
> |
self.dbspath=dataset+'/datatier/'+owner |
| 30 |
> |
|
| 31 |
> |
self.api = dbsCgiApi.DbsCgiApi(cgiUrl="http://cern.ch/cms-dbs/cgi-bin") |
| 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 |
| 38 |
> |
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 |
|
|
| 45 |
+ |
return parent |
| 46 |
|
|
| 47 |
|
# #################################### |
| 48 |
|
def getDatasetContents(self): |
| 49 |
|
""" |
| 50 |
|
query DBS to get event collections |
| 51 |
|
""" |
| 52 |
< |
print "DBSInfo: ---> getDatasetContents : "+self.dataset+"/"+self.owner |
| 53 |
< |
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) |
| 52 |
> |
|
| 53 |
> |
fileBlockList = self.api.getDatasetContents(self.dbspath) |
| 54 |
|
|
| 55 |
|
## get the fileblock and event collections |
| 56 |
|
nevtsbyblock= {} |
| 57 |
< |
print "DBSInfo: fileblocks are: %s"%handler.fileblocksList |
| 57 |
> |
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 |
|
|
| 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 |
| 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 |
< |
|
| 71 |
> |
|
| 72 |
|
return nevtsbyblock |
| 73 |
|
|
| 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 |
– |
|
