| 1 |
|
#!/usr/bin/env python2 |
| 2 |
|
import sys, os, string, re |
| 3 |
– |
import urllib |
| 4 |
– |
import urllib2 |
| 3 |
|
from DBSInfo import * |
| 4 |
|
|
| 5 |
+ |
|
| 6 |
+ |
# #################################### |
| 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 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 DataDiscoveryError: |
| 30 |
< |
def __init__(self): |
| 31 |
< |
print '\nERROR accessing Data Discovery\n' |
| 32 |
< |
pass |
| 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 |
| 42 |
|
def __init__(self, owner, dataset, dataTiers, cfg_params): |
| 43 |
|
|
| 44 |
|
# Attributes |
| 45 |
< |
self.owner = owner |
| 20 |
< |
self.dataset = dataset |
| 45 |
> |
self.dbsdataset='/'+dataset+'/datatier/'+owner |
| 46 |
|
self.dataTiers = dataTiers |
| 47 |
|
self.cfg_params = cfg_params |
| 48 |
|
|
| 49 |
< |
self.dataset_owner = {} # DBS output: map dataset-owner for all data |
| 49 |
> |
self.dbspaths= [] # DBS output: list of dbspaths for all data |
| 50 |
|
self.allblocks = [] # DBS output: list of map fileblocks-totevts for all dataset-owners |
| 51 |
< |
self.blocksinfo = {} # DBS output: map fileblocks-totevts for the primary block, used internally to this class |
| 51 |
> |
self.blocksinfo = {} # DBS output: map fileblocks-totevts for the primary block, used internally to this class |
| 52 |
|
#DBS output: max events computed by method getMaxEvents |
| 53 |
|
|
| 54 |
|
# #################################### |
| 56 |
|
""" |
| 57 |
|
Contact DBS |
| 58 |
|
""" |
| 59 |
< |
parents = {} |
| 59 |
> |
parents = [] |
| 60 |
|
parentsblocksinfo = {} |
| 36 |
– |
self.dataset_owner[self.owner]=self.dataset # add the map dataset-owner |
| 61 |
|
|
| 62 |
< |
dbs=DBSInfo(self.owner,self.dataset,self.dataTiers) |
| 62 |
> |
## add the PU among the required data tiers if the Digi are requested |
| 63 |
> |
if (self.dataTiers.count('Digi')>0) & (self.dataTiers.count('PU')<=0) : |
| 64 |
> |
self.dataTiers.append('PU') |
| 65 |
> |
|
| 66 |
> |
## get info about the requested dataset |
| 67 |
> |
dbs=DBSInfo(self.dbsdataset,self.dataTiers) |
| 68 |
|
try: |
| 69 |
|
self.blocksinfo=dbs.getDatasetContents() |
| 70 |
< |
except: |
| 71 |
< |
raise DataDiscoveryError |
| 70 |
> |
except DBSError, ex: |
| 71 |
> |
raise DataDiscoveryError(ex.getErrorMessage()) |
| 72 |
> |
|
| 73 |
> |
if len(self.blocksinfo)<=0: |
| 74 |
> |
msg="\nData %s do not exist in DBS! \n Check the dataset/owner variables in crab.cfg !"%self.dbsdataset |
| 75 |
> |
raise NotExistingDatasetError(msg) |
| 76 |
> |
|
| 77 |
> |
currentdatatier=string.split(self.blocksinfo.keys()[0],'/')[2] |
| 78 |
> |
fakedatatier=string.split(self.dbsdataset,'/')[2] |
| 79 |
> |
currentdbsdataset=string.replace(self.dbsdataset, fakedatatier, currentdatatier) |
| 80 |
> |
|
| 81 |
> |
self.dbspaths.append(currentdbsdataset) # add the requested dbspath |
| 82 |
|
|
| 83 |
+ |
## get info about the parents |
| 84 |
|
try: |
| 85 |
|
parents=dbs.getDatasetProvenance() |
| 86 |
< |
except: |
| 87 |
< |
raise DataDiscoveryError |
| 86 |
> |
except DBSInvalidDataTierError, ex: |
| 87 |
> |
msg=ex.getErrorMessage()+' \n Check the data_tier variable in crab.cfg !\n' |
| 88 |
> |
raise DataDiscoveryError(msg) |
| 89 |
> |
except DBSError, ex: |
| 90 |
> |
raise DataDiscoveryError(ex.getErrorMessage()) |
| 91 |
> |
|
| 92 |
> |
## check that the user asks for parent Data Tier really existing in the DBS provenance |
| 93 |
> |
self.checkParentDataTier(parents, self.dataTiers, currentdbsdataset) |
| 94 |
|
|
| 95 |
|
## for each parent get the corresponding fileblocks |
| 96 |
|
for aparent in parents: |
| 97 |
< |
## fill the map dataset-owner for the parents |
| 98 |
< |
pdataset=string.split(aparent,'/')[1] |
| 99 |
< |
powner=string.split(aparent,'/')[3] |
| 100 |
< |
self.dataset_owner[powner]=pdataset |
| 101 |
< |
## get the fileblocks of the parents |
| 102 |
< |
pdbs=DBSInfo(powner,pdataset,[]) |
| 103 |
< |
parentsblocksinfo=pdbs.getDatasetContents() |
| 97 |
> |
## fill a list of dbspaths |
| 98 |
> |
parentdbsdataset=aparent.getDatasetPath() |
| 99 |
> |
self.dbspaths.append(parentdbsdataset) |
| 100 |
> |
pdbs=DBSInfo(parentdbsdataset,[]) |
| 101 |
> |
try: |
| 102 |
> |
parentsblocksinfo=pdbs.getDatasetContents() |
| 103 |
> |
except DBSError, ex: |
| 104 |
> |
raise DataDiscoveryError(ex.getErrorMessage()) |
| 105 |
> |
|
| 106 |
|
self.allblocks.append(parentsblocksinfo.keys()) # add parent fileblocksinfo |
| 107 |
|
|
| 108 |
|
## all the required blocks |
| 109 |
< |
self.allblocks.append(self.blocksinfo.keys()) # add also the primary fileblocksinfo |
| 109 |
> |
self.allblocks.append(self.blocksinfo.keys()) # add also the current fileblocksinfo |
| 110 |
> |
|
| 111 |
> |
|
| 112 |
> |
# ################################################# |
| 113 |
> |
def checkParentDataTier(self, parents, user_datatiers, currentdbsdataset ): |
| 114 |
> |
""" |
| 115 |
> |
check that the data tiers requested by the user really exists in the provenance of the given dataset |
| 116 |
> |
""" |
| 117 |
> |
|
| 118 |
> |
current_datatier=string.split(currentdbsdataset,'/')[2] |
| 119 |
> |
|
| 120 |
> |
parent_datatypes=[] |
| 121 |
> |
for aparent in parents: |
| 122 |
> |
parent_datatypes.append(aparent.getDataType()) |
| 123 |
> |
|
| 124 |
> |
for datatier in user_datatiers: |
| 125 |
> |
if parent_datatypes.count(datatier)<=0: |
| 126 |
> |
# the current datatier is not supposed to be in the provenance |
| 127 |
> |
if not (datatier == current_datatier): |
| 128 |
> |
msg="\nData %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 !"%(currentdbsdataset,datatier) |
| 129 |
> |
raise NoDataTierinProvenanceError(msg) |
| 130 |
|
|
| 131 |
|
|
| 132 |
|
# ################################################# |
| 142 |
|
return nevts |
| 143 |
|
|
| 144 |
|
# ################################################# |
| 145 |
< |
def getDatasetOwnerPairs(self): |
| 145 |
> |
def getDBSPaths(self): |
| 146 |
|
""" |
| 147 |
< |
list all required dataset-owner pairs |
| 147 |
> |
list the DBSpaths for all required data |
| 148 |
|
""" |
| 149 |
< |
return self.dataset_owner |
| 149 |
> |
return self.dbspaths |
| 150 |
|
|
| 151 |
|
# ################################################# |
| 152 |
|
def getEVC(self): |
