| 1 |
< |
#!/usr/bin/env python2 |
| 1 |
> |
#!/usr/bin/env python |
| 2 |
|
import sys, os, string, re |
| 3 |
|
from DBSInfo import * |
| 4 |
|
|
| 5 |
+ |
|
| 6 |
|
# #################################### |
| 7 |
< |
class DataDiscoveryError: |
| 8 |
< |
def __init__(self): |
| 9 |
< |
print '\nERROR accessing Data Discovery\n' |
| 7 |
> |
class DataDiscoveryError(exceptions.Exception): |
| 8 |
> |
def __init__(self, errorMessage): |
| 9 |
> |
args=errorMessage |
| 10 |
> |
exceptions.Exception.__init__(self, args) |
| 11 |
|
pass |
| 12 |
+ |
|
| 13 |
+ |
def getErrorMessage(self): |
| 14 |
+ |
""" Return exception error """ |
| 15 |
+ |
return "%s" % (self.args) |
| 16 |
+ |
|
| 17 |
|
# #################################### |
| 18 |
< |
class DatasetContentsError: |
| 19 |
< |
def __init__(self): |
| 20 |
< |
print '\nERROR accessing Data Discovery : getDatasetContents\n' |
| 18 |
> |
class NotExistingDatasetError(exceptions.Exception): |
| 19 |
> |
def __init__(self, errorMessage): |
| 20 |
> |
args=errorMessage |
| 21 |
> |
exceptions.Exception.__init__(self, args) |
| 22 |
|
pass |
| 23 |
|
|
| 24 |
+ |
def getErrorMessage(self): |
| 25 |
+ |
""" Return exception error """ |
| 26 |
+ |
return "%s" % (self.args) |
| 27 |
+ |
|
| 28 |
|
# #################################### |
| 29 |
< |
class DatasetProvenanceError: |
| 30 |
< |
def __init__(self): |
| 31 |
< |
print '\nERROR accessing Data Discovery : getDatasetProvenance\n' |
| 29 |
> |
class NoDataTierinProvenanceError(exceptions.Exception): |
| 30 |
> |
def __init__(self, errorMessage): |
| 31 |
> |
args=errorMessage |
| 32 |
> |
exceptions.Exception.__init__(self, args) |
| 33 |
|
pass |
| 34 |
|
|
| 35 |
+ |
def getErrorMessage(self): |
| 36 |
+ |
""" Return exception error """ |
| 37 |
+ |
return "%s" % (self.args) |
| 38 |
+ |
|
| 39 |
|
# #################################### |
| 40 |
|
# class to find and extact info from published data |
| 41 |
|
class DataDiscovery: |
| 42 |
|
def __init__(self, owner, dataset, dataTiers, cfg_params): |
| 43 |
|
|
| 44 |
|
# Attributes |
| 45 |
< |
self.dbsdataset=dataset+'/datatier/'+owner |
| 45 |
> |
self.owner = owner |
| 46 |
> |
self.dataset = dataset |
| 47 |
|
self.dataTiers = dataTiers |
| 48 |
|
self.cfg_params = cfg_params |
| 49 |
|
|
| 50 |
< |
self.dbspaths= [] # DBS output: list of dbspaths for all data |
| 50 |
> |
self.dbspaths= [] # DBS output: list of dbspaths for all data |
| 51 |
|
self.allblocks = [] # DBS output: list of map fileblocks-totevts for all dataset-owners |
| 52 |
< |
self.blocksinfo = {} # DBS output: map fileblocks-totevts for the primary block, used internally to this class |
| 52 |
> |
self.blocksinfo = {} # DBS output: map fileblocks-totevts for the primary block, used internally to this class |
| 53 |
|
#DBS output: max events computed by method getMaxEvents |
| 54 |
|
|
| 55 |
|
# #################################### |
| 57 |
|
""" |
| 58 |
|
Contact DBS |
| 59 |
|
""" |
| 42 |
– |
parents = [] |
| 43 |
– |
parentsblocksinfo = {} |
| 44 |
– |
self.dbspaths.append("/"+self.dbsdataset) # add the primary dbspath |
| 45 |
– |
# it might be replaced if one get from DBSAPI the primary dbspath as well |
| 60 |
|
|
| 61 |
< |
dbs=DBSInfo(self.dbsdataset,self.dataTiers) |
| 61 |
> |
## 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() |
| 67 |
|
try: |
| 68 |
< |
self.blocksinfo=dbs.getDatasetContents() |
| 69 |
< |
except dbs.DBSError: |
| 70 |
< |
raise DataDiscoveryError |
| 68 |
> |
self.datasets = dbs.getMatchingDatasets(self.owner, self.dataset) |
| 69 |
> |
except DBSError, ex: |
| 70 |
> |
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 |
< |
parents=dbs.getDatasetProvenance() |
| 77 |
< |
except: |
| 78 |
< |
raise DataDiscoveryError |
| 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) |
| 86 |
|
|
| 57 |
– |
## for each parent get the corresponding fileblocks |
| 58 |
– |
for aparent in parents: |
| 59 |
– |
## fill the map dataset-owner for the parents |
| 60 |
– |
#pdataset=string.split(aparent,'/')[1] |
| 61 |
– |
#powner=string.split(aparent,'/')[3] |
| 62 |
– |
#self.dataset_owner[powner]=pdataset |
| 63 |
– |
## instead of the map dataset-owner use the dbspaths |
| 64 |
– |
parentdbsdataset=aparent.getDatasetPath() |
| 65 |
– |
self.dbspaths.append(parentdbsdataset) |
| 66 |
– |
#self.dbspaths.append(aparent) |
| 67 |
– |
## get the fileblocks of the parents : FIXME remove the first / in the path |
| 68 |
– |
pdbs=DBSInfo(parentdbsdataset[1:-1],[]) |
| 69 |
– |
try: |
| 70 |
– |
parentsblocksinfo=pdbs.getDatasetContents() |
| 71 |
– |
except: |
| 72 |
– |
raise DataDiscoveryError |
| 87 |
|
|
| 88 |
< |
self.allblocks.append(parentsblocksinfo.keys()) # add parent fileblocksinfo |
| 88 |
> |
## get info about the parents |
| 89 |
> |
try: |
| 90 |
> |
parents=dbs.getDatasetProvenance(self.dbsdataset, self.dataTiers) |
| 91 |
> |
except DBSInvalidDataTierError, ex: |
| 92 |
> |
msg=ex.getErrorMessage()+' \n Check the data_tier variable in crab.cfg !\n' |
| 93 |
> |
raise DataDiscoveryError(msg) |
| 94 |
> |
except DBSError, ex: |
| 95 |
> |
raise DataDiscoveryError(ex.getErrorMessage()) |
| 96 |
> |
|
| 97 |
> |
## check that the user asks for parent Data Tier really existing in the DBS provenance |
| 98 |
> |
self.checkParentDataTier(parents, self.dataTiers) |
| 99 |
> |
|
| 100 |
> |
## for each parent get the corresponding fileblocks |
| 101 |
> |
try: |
| 102 |
> |
for p in parents: |
| 103 |
> |
## fill a list of dbspaths |
| 104 |
> |
parentPath = p.get('parent').get('datasetPathName') |
| 105 |
> |
self.dbspaths.append (parentPath) |
| 106 |
> |
parentBlocks = dbs.getDatasetContents (parentPath) |
| 107 |
> |
self.allblocks.append (parentBlocks.keys ()) # add parent fileblocksinfo |
| 108 |
> |
except DBSError, ex: |
| 109 |
> |
raise DataDiscoveryError(ex.getErrorMessage()) |
| 110 |
|
|
| 111 |
< |
## all the required blocks |
| 112 |
< |
self.allblocks.append(self.blocksinfo.keys()) # add also the primary fileblocksinfo |
| 111 |
> |
# ################################################# |
| 112 |
> |
def checkParentDataTier(self, parents, dataTiers): |
| 113 |
> |
""" |
| 114 |
> |
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) |
| 123 |
|
|
| 124 |
|
|
| 125 |
|
# ################################################# |
| 126 |
|
def getMaxEvents(self): |
| 127 |
|
""" |
| 128 |
< |
max events of the primary dataset-owner |
| 128 |
> |
max events of the primary dataset-owner |
| 129 |
|
""" |
| 130 |
|
## loop over the fileblocks of the primary dataset-owner |
| 131 |
|
nevts=0 |
| 132 |
|
for blockevts in self.blocksinfo.values(): |
| 133 |
< |
nevts=nevts+blockevts |
| 133 |
> |
nevts=nevts+blockevts |
| 134 |
|
|
| 135 |
|
return nevts |
| 136 |
|
|
| 137 |
|
# ################################################# |
| 93 |
– |
def getDatasetOwnerPairs(self): |
| 94 |
– |
""" |
| 95 |
– |
list all required dataset-owner pairs |
| 96 |
– |
""" |
| 97 |
– |
return self.dataset_owner |
| 98 |
– |
# ################################################# |
| 138 |
|
def getDBSPaths(self): |
| 139 |
|
""" |
| 140 |
< |
list the DBSpaths for all required data |
| 140 |
> |
list the DBSpaths for all required data |
| 141 |
|
""" |
| 142 |
|
return self.dbspaths |
| 143 |
|
|
| 144 |
|
# ################################################# |
| 145 |
|
def getEVC(self): |
| 146 |
|
""" |
| 147 |
< |
list the event collections structure by fileblock |
| 147 |
> |
list the event collections structure by fileblock |
| 148 |
|
""" |
| 149 |
|
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" |
| 152 |
|
# ################################################# |
| 153 |
|
def getFileBlocks(self): |
| 154 |
|
""" |
| 155 |
< |
fileblocks for all required dataset-owners |
| 155 |
> |
fileblocks for all required dataset-owners |
| 156 |
|
""" |
| 157 |
|
return self.allblocks |
| 158 |
|
|
| 159 |
|
######################################################################## |
| 121 |
– |
|
| 122 |
– |
|
