1 |
gutsche |
1.6 |
#!/usr/bin/env python
|
2 |
afanfani |
1.1 |
from DBSInfo import *
|
3 |
|
|
|
4 |
afanfani |
1.3 |
|
5 |
afanfani |
1.1 |
# ####################################
|
6 |
afanfani |
1.3 |
class DataDiscoveryError(exceptions.Exception):
|
7 |
slacapra |
1.7 |
def __init__(self, errorMessage):
|
8 |
slacapra |
1.14 |
exceptions.Exception.__init__(self, self.args)
|
9 |
|
|
self.args=errorMessage
|
10 |
slacapra |
1.7 |
pass
|
11 |
|
|
|
12 |
|
|
def getErrorMessage(self):
|
13 |
|
|
""" Return exception error """
|
14 |
|
|
return "%s" % (self.args)
|
15 |
afanfani |
1.3 |
|
16 |
afanfani |
1.1 |
# ####################################
|
17 |
afanfani |
1.3 |
class NotExistingDatasetError(exceptions.Exception):
|
18 |
slacapra |
1.7 |
def __init__(self, errorMessage):
|
19 |
slacapra |
1.14 |
exceptions.Exception.__init__(self, self.args)
|
20 |
|
|
self.args=errorMessage
|
21 |
slacapra |
1.7 |
pass
|
22 |
|
|
|
23 |
|
|
def getErrorMessage(self):
|
24 |
|
|
""" Return exception error """
|
25 |
|
|
return "%s" % (self.args)
|
26 |
afanfani |
1.1 |
|
27 |
|
|
# ####################################
|
28 |
afanfani |
1.3 |
class NoDataTierinProvenanceError(exceptions.Exception):
|
29 |
slacapra |
1.7 |
def __init__(self, errorMessage):
|
30 |
slacapra |
1.14 |
exceptions.Exception.__init__(self, self.args)
|
31 |
|
|
self.args=errorMessage
|
32 |
slacapra |
1.7 |
pass
|
33 |
|
|
|
34 |
|
|
def getErrorMessage(self):
|
35 |
|
|
""" Return exception error """
|
36 |
|
|
return "%s" % (self.args)
|
37 |
afanfani |
1.1 |
|
38 |
|
|
# ####################################
|
39 |
|
|
# class to find and extact info from published data
|
40 |
|
|
class DataDiscovery:
|
41 |
slacapra |
1.11 |
def __init__(self, datasetPath, dataTiers, cfg_params):
|
42 |
afanfani |
1.1 |
|
43 |
|
|
# Attributes
|
44 |
slacapra |
1.11 |
self.datasetPath = datasetPath
|
45 |
afanfani |
1.1 |
self.dataTiers = dataTiers
|
46 |
|
|
self.cfg_params = cfg_params
|
47 |
|
|
|
48 |
slacapra |
1.11 |
self.eventsPerBlock = {} # DBS output: map fileblocks-events for collection
|
49 |
|
|
self.eventsPerFile = {} # DBS output: map files-events
|
50 |
|
|
self.blocksinfo = {} # DBS output: map fileblocks-files
|
51 |
afanfani |
1.1 |
#DBS output: max events computed by method getMaxEvents
|
52 |
|
|
|
53 |
|
|
# ####################################
|
54 |
|
|
def fetchDBSInfo(self):
|
55 |
|
|
"""
|
56 |
|
|
Contact DBS
|
57 |
|
|
"""
|
58 |
|
|
|
59 |
slacapra |
1.11 |
## get DBS URL
|
60 |
|
|
try:
|
61 |
|
|
dbs_url=self.cfg_params['CMSSW.dbs_url']
|
62 |
|
|
except KeyError:
|
63 |
|
|
dbs_url="http://cmsdoc.cern.ch/cms/test/aprom/DBS/CGIServer/prodquery"
|
64 |
afanfani |
1.3 |
|
65 |
slacapra |
1.11 |
## get info about the requested dataset
|
66 |
|
|
try:
|
67 |
|
|
dbs_instance=self.cfg_params['CMSSW.dbs_instance']
|
68 |
|
|
except KeyError:
|
69 |
gutsche |
1.12 |
dbs_instance="MCGlobal/Writer"
|
70 |
slacapra |
1.11 |
|
71 |
|
|
dbs = DBSInfo(dbs_url, dbs_instance)
|
72 |
afanfani |
1.5 |
try:
|
73 |
slacapra |
1.11 |
self.datasets = dbs.getMatchingDatasets(self.datasetPath)
|
74 |
|
|
except dbsCgiApi.DbsCgiExecutionError, msg:
|
75 |
|
|
raise DataDiscoveryError(msg)
|
76 |
slacapra |
1.13 |
except DBSError, msg:
|
77 |
|
|
raise DataDiscoveryError(msg)
|
78 |
slacapra |
1.11 |
|
79 |
afanfani |
1.4 |
if len(self.datasets) == 0:
|
80 |
slacapra |
1.11 |
raise DataDiscoveryError("DatasetPath=%s unknown to DBS" %self.datasetPath)
|
81 |
afanfani |
1.4 |
if len(self.datasets) > 1:
|
82 |
slacapra |
1.11 |
raise DataDiscoveryError("DatasetPath=%s is ambiguous" %self.datasetPath)
|
83 |
|
|
|
84 |
afanfani |
1.1 |
try:
|
85 |
slacapra |
1.7 |
self.dbsdataset = self.datasets[0].get('datasetPathName')
|
86 |
slacapra |
1.11 |
|
87 |
|
|
self.eventsPerBlock = dbs.getEventsPerBlock(self.dbsdataset)
|
88 |
|
|
self.blocksinfo = dbs.getDatasetFileBlocks(self.dbsdataset)
|
89 |
|
|
self.eventsPerFile = dbs.getEventsPerFile(self.dbsdataset)
|
90 |
afanfani |
1.3 |
except DBSError, ex:
|
91 |
slacapra |
1.7 |
raise DataDiscoveryError(ex.getErrorMessage())
|
92 |
afanfani |
1.3 |
|
93 |
slacapra |
1.11 |
if len(self.eventsPerBlock) <= 0:
|
94 |
|
|
raise NotExistingDatasetError (("\nNo data for %s in DBS\nPlease check"
|
95 |
|
|
+ " dataset path variables in crab.cfg")
|
96 |
|
|
% self.dbsdataset)
|
97 |
afanfani |
1.1 |
|
98 |
|
|
|
99 |
|
|
# #################################################
|
100 |
|
|
def getMaxEvents(self):
|
101 |
|
|
"""
|
102 |
slacapra |
1.11 |
max events
|
103 |
afanfani |
1.1 |
"""
|
104 |
slacapra |
1.11 |
## loop over the event collections
|
105 |
afanfani |
1.1 |
nevts=0
|
106 |
slacapra |
1.11 |
for evc_evts in self.eventsPerBlock.values():
|
107 |
|
|
nevts=nevts+evc_evts
|
108 |
afanfani |
1.1 |
|
109 |
|
|
return nevts
|
110 |
|
|
|
111 |
|
|
# #################################################
|
112 |
slacapra |
1.11 |
def getEventsPerBlock(self):
|
113 |
afanfani |
1.1 |
"""
|
114 |
slacapra |
1.11 |
list the event collections structure by fileblock
|
115 |
afanfani |
1.1 |
"""
|
116 |
slacapra |
1.11 |
return self.eventsPerBlock
|
117 |
afanfani |
1.1 |
|
118 |
|
|
# #################################################
|
119 |
slacapra |
1.11 |
def getEventsPerFile(self):
|
120 |
afanfani |
1.1 |
"""
|
121 |
slacapra |
1.11 |
list the event collections structure by file
|
122 |
afanfani |
1.1 |
"""
|
123 |
slacapra |
1.11 |
return self.eventsPerFile
|
124 |
afanfani |
1.1 |
|
125 |
|
|
# #################################################
|
126 |
slacapra |
1.11 |
def getFiles(self):
|
127 |
afanfani |
1.1 |
"""
|
128 |
slacapra |
1.11 |
return files grouped by fileblock
|
129 |
afanfani |
1.1 |
"""
|
130 |
slacapra |
1.11 |
return self.blocksinfo
|
131 |
afanfani |
1.1 |
|
132 |
|
|
########################################################################
|