1 |
gutsche |
1.6 |
#!/usr/bin/env python
|
2 |
slacapra |
1.18 |
import exceptions
|
3 |
|
|
import DBSAPI.dbsApi
|
4 |
|
|
from DBSAPI.dbsApiException import *
|
5 |
|
|
import common
|
6 |
|
|
from crab_util import *
|
7 |
afanfani |
1.1 |
|
8 |
afanfani |
1.3 |
|
9 |
slacapra |
1.18 |
# #######################################
|
10 |
|
|
class DBSError(exceptions.Exception):
|
11 |
|
|
def __init__(self, errorName, errorMessage):
|
12 |
|
|
args='\nERROR DBS %s : %s \n'%(errorName,errorMessage)
|
13 |
|
|
exceptions.Exception.__init__(self, args)
|
14 |
|
|
pass
|
15 |
|
|
|
16 |
|
|
def getErrorMessage(self):
|
17 |
|
|
""" Return error message """
|
18 |
|
|
return "%s" % (self.args)
|
19 |
|
|
|
20 |
|
|
# #######################################
|
21 |
|
|
class DBSInvalidDataTierError(exceptions.Exception):
|
22 |
|
|
def __init__(self, errorName, errorMessage):
|
23 |
|
|
args='\nERROR DBS %s : %s \n'%(errorName,errorMessage)
|
24 |
|
|
exceptions.Exception.__init__(self, args)
|
25 |
|
|
pass
|
26 |
|
|
|
27 |
|
|
def getErrorMessage(self):
|
28 |
|
|
""" Return error message """
|
29 |
|
|
return "%s" % (self.args)
|
30 |
|
|
|
31 |
|
|
# #######################################
|
32 |
|
|
class DBSInfoError:
|
33 |
|
|
def __init__(self, url):
|
34 |
|
|
print '\nERROR accessing DBS url : '+url+'\n'
|
35 |
|
|
pass
|
36 |
|
|
|
37 |
afanfani |
1.1 |
# ####################################
|
38 |
afanfani |
1.3 |
class DataDiscoveryError(exceptions.Exception):
|
39 |
slacapra |
1.7 |
def __init__(self, errorMessage):
|
40 |
gutsche |
1.15 |
self.args=errorMessage
|
41 |
slacapra |
1.14 |
exceptions.Exception.__init__(self, self.args)
|
42 |
slacapra |
1.7 |
pass
|
43 |
|
|
|
44 |
|
|
def getErrorMessage(self):
|
45 |
|
|
""" Return exception error """
|
46 |
|
|
return "%s" % (self.args)
|
47 |
afanfani |
1.3 |
|
48 |
afanfani |
1.1 |
# ####################################
|
49 |
afanfani |
1.3 |
class NotExistingDatasetError(exceptions.Exception):
|
50 |
slacapra |
1.7 |
def __init__(self, errorMessage):
|
51 |
gutsche |
1.15 |
self.args=errorMessage
|
52 |
slacapra |
1.14 |
exceptions.Exception.__init__(self, self.args)
|
53 |
slacapra |
1.7 |
pass
|
54 |
|
|
|
55 |
|
|
def getErrorMessage(self):
|
56 |
|
|
""" Return exception error """
|
57 |
|
|
return "%s" % (self.args)
|
58 |
afanfani |
1.1 |
|
59 |
|
|
# ####################################
|
60 |
afanfani |
1.3 |
class NoDataTierinProvenanceError(exceptions.Exception):
|
61 |
slacapra |
1.7 |
def __init__(self, errorMessage):
|
62 |
gutsche |
1.15 |
self.args=errorMessage
|
63 |
slacapra |
1.14 |
exceptions.Exception.__init__(self, self.args)
|
64 |
slacapra |
1.7 |
pass
|
65 |
|
|
|
66 |
|
|
def getErrorMessage(self):
|
67 |
|
|
""" Return exception error """
|
68 |
|
|
return "%s" % (self.args)
|
69 |
afanfani |
1.1 |
|
70 |
|
|
# ####################################
|
71 |
|
|
# class to find and extact info from published data
|
72 |
|
|
class DataDiscovery:
|
73 |
gutsche |
1.15 |
def __init__(self, datasetPath, cfg_params):
|
74 |
afanfani |
1.1 |
|
75 |
slacapra |
1.18 |
# Attributes
|
76 |
slacapra |
1.11 |
self.datasetPath = datasetPath
|
77 |
afanfani |
1.1 |
self.cfg_params = cfg_params
|
78 |
|
|
|
79 |
slacapra |
1.11 |
self.eventsPerBlock = {} # DBS output: map fileblocks-events for collection
|
80 |
|
|
self.eventsPerFile = {} # DBS output: map files-events
|
81 |
slacapra |
1.18 |
self.blocksinfo = {} # DBS output: map fileblocks-files
|
82 |
|
|
self.maxEvents = 0 # DBS output: max events
|
83 |
spiga |
1.21 |
self.parent = {} # DBS output: max events
|
84 |
afanfani |
1.1 |
|
85 |
|
|
# ####################################
|
86 |
|
|
def fetchDBSInfo(self):
|
87 |
|
|
"""
|
88 |
|
|
Contact DBS
|
89 |
|
|
"""
|
90 |
|
|
|
91 |
slacapra |
1.11 |
## get DBS URL
|
92 |
slacapra |
1.19 |
dbs_url="http://cmsdbsprod.cern.ch/cms_dbs_prod_global/servlet/DBSServlet"
|
93 |
|
|
if (self.cfg_params.has_key('CMSSW.dbs_url')):
|
94 |
slacapra |
1.11 |
dbs_url=self.cfg_params['CMSSW.dbs_url']
|
95 |
afanfani |
1.3 |
|
96 |
slacapra |
1.18 |
common.logger.debug(3,"Accessing DBS at: "+dbs_url)
|
97 |
|
|
|
98 |
|
|
## check if runs are selected
|
99 |
slacapra |
1.19 |
runselection = []
|
100 |
|
|
if (self.cfg_params.has_key('CMSSW.runselection')):
|
101 |
slacapra |
1.18 |
runselection = parseRange2(self.cfg_params['CMSSW.runselection'])
|
102 |
|
|
|
103 |
afanfani |
1.20 |
common.logger.debug(6,"runselection is: %s"%runselection)
|
104 |
slacapra |
1.18 |
## service API
|
105 |
|
|
args = {}
|
106 |
|
|
args['url'] = dbs_url
|
107 |
|
|
args['level'] = 'CRITICAL'
|
108 |
|
|
|
109 |
spiga |
1.21 |
## check if has been requested to use the parent info
|
110 |
|
|
if (self.cfg_params.has_key('CMSSW.runselection')):
|
111 |
|
|
runselection = parseRange2(self.cfg_params['CMSSW.runselection'])
|
112 |
|
|
|
113 |
|
|
useParent = self.cfg_params.get('CMSSW.use_parent',False)
|
114 |
|
|
|
115 |
|
|
allowedRetriveValue = [
|
116 |
|
|
'retrive_child',
|
117 |
|
|
'retrive_block',
|
118 |
|
|
'retrive_lumi',
|
119 |
|
|
'retrive_run'
|
120 |
|
|
]
|
121 |
|
|
if useParent: allowedRetriveValue.append('retrive_parent')
|
122 |
|
|
common.logger.debug(5,"Set of input parameters used for DBS query : \n"+str(allowedRetriveValue))
|
123 |
|
|
common.logger.write("Set of input parameters used for DBS query : \n"+str(allowedRetriveValue))
|
124 |
slacapra |
1.18 |
api = DBSAPI.dbsApi.DbsApi(args)
|
125 |
afanfani |
1.5 |
try:
|
126 |
slacapra |
1.18 |
if len(runselection) <= 0 :
|
127 |
spiga |
1.21 |
files = api.listFiles(path=self.datasetPath,retriveList=allowedRetriveValue)
|
128 |
slacapra |
1.18 |
else :
|
129 |
afanfani |
1.20 |
files=[]
|
130 |
|
|
for arun in runselection:
|
131 |
|
|
try:
|
132 |
spiga |
1.21 |
filesinrun = api.listFiles(path=self.datasetPath,retriveList=allowedRetriveValue,runNumber=arun)
|
133 |
afanfani |
1.20 |
files.extend(filesinrun)
|
134 |
|
|
except:
|
135 |
|
|
msg="WARNING: problem extracting info from DBS for run %s "%arun
|
136 |
|
|
common.logger.message(msg)
|
137 |
|
|
pass
|
138 |
|
|
|
139 |
slacapra |
1.18 |
except DbsBadRequest, msg:
|
140 |
slacapra |
1.11 |
raise DataDiscoveryError(msg)
|
141 |
slacapra |
1.13 |
except DBSError, msg:
|
142 |
|
|
raise DataDiscoveryError(msg)
|
143 |
slacapra |
1.11 |
|
144 |
slacapra |
1.18 |
# parse files and fill arrays
|
145 |
|
|
for file in files :
|
146 |
spiga |
1.21 |
parList = []
|
147 |
slacapra |
1.18 |
filename = file['LogicalFileName']
|
148 |
spiga |
1.21 |
# asked retry the list of parent for the given child
|
149 |
|
|
if useParent: parList = [x['LogicalFileName'] for x in file['ParentList']]
|
150 |
|
|
self.parent[filename] = parList
|
151 |
slacapra |
1.18 |
if filename.find('.dat') < 0 :
|
152 |
|
|
fileblock = file['Block']['Name']
|
153 |
|
|
events = file['NumberOfEvents']
|
154 |
afanfani |
1.20 |
# number of events per block
|
155 |
|
|
if fileblock in self.eventsPerBlock.keys() :
|
156 |
|
|
self.eventsPerBlock[fileblock] += events
|
157 |
|
|
else :
|
158 |
|
|
self.eventsPerBlock[fileblock] = events
|
159 |
|
|
# number of events per file
|
160 |
|
|
self.eventsPerFile[filename] = events
|
161 |
|
|
|
162 |
|
|
# number of events per block
|
163 |
|
|
if fileblock in self.blocksinfo.keys() :
|
164 |
|
|
self.blocksinfo[fileblock].append(filename)
|
165 |
slacapra |
1.18 |
else :
|
166 |
afanfani |
1.20 |
self.blocksinfo[fileblock] = [filename]
|
167 |
slacapra |
1.18 |
|
168 |
afanfani |
1.20 |
# total number of events
|
169 |
|
|
self.maxEvents += events
|
170 |
slacapra |
1.11 |
|
171 |
slacapra |
1.18 |
for block in self.eventsPerBlock.keys() :
|
172 |
|
|
common.logger.debug(6,"DBSInfo: total nevts %i in block %s "%(self.eventsPerBlock[block],block))
|
173 |
slacapra |
1.11 |
|
174 |
|
|
if len(self.eventsPerBlock) <= 0:
|
175 |
slacapra |
1.18 |
raise NotExistingDatasetError(("\nNo data for %s in DBS\nPlease check"
|
176 |
slacapra |
1.11 |
+ " dataset path variables in crab.cfg")
|
177 |
slacapra |
1.18 |
% self.datasetPath)
|
178 |
afanfani |
1.1 |
|
179 |
|
|
|
180 |
|
|
# #################################################
|
181 |
|
|
def getMaxEvents(self):
|
182 |
|
|
"""
|
183 |
slacapra |
1.11 |
max events
|
184 |
afanfani |
1.1 |
"""
|
185 |
slacapra |
1.18 |
return self.maxEvents
|
186 |
afanfani |
1.1 |
|
187 |
|
|
# #################################################
|
188 |
slacapra |
1.11 |
def getEventsPerBlock(self):
|
189 |
afanfani |
1.1 |
"""
|
190 |
slacapra |
1.11 |
list the event collections structure by fileblock
|
191 |
afanfani |
1.1 |
"""
|
192 |
slacapra |
1.11 |
return self.eventsPerBlock
|
193 |
afanfani |
1.1 |
|
194 |
|
|
# #################################################
|
195 |
slacapra |
1.11 |
def getEventsPerFile(self):
|
196 |
afanfani |
1.1 |
"""
|
197 |
slacapra |
1.11 |
list the event collections structure by file
|
198 |
afanfani |
1.1 |
"""
|
199 |
slacapra |
1.11 |
return self.eventsPerFile
|
200 |
afanfani |
1.1 |
|
201 |
|
|
# #################################################
|
202 |
slacapra |
1.11 |
def getFiles(self):
|
203 |
afanfani |
1.1 |
"""
|
204 |
slacapra |
1.11 |
return files grouped by fileblock
|
205 |
afanfani |
1.1 |
"""
|
206 |
slacapra |
1.11 |
return self.blocksinfo
|
207 |
afanfani |
1.1 |
|
208 |
spiga |
1.21 |
# #################################################
|
209 |
|
|
def getParent(self):
|
210 |
|
|
"""
|
211 |
|
|
return parent grouped by file
|
212 |
|
|
"""
|
213 |
|
|
return self.parent
|
214 |
|
|
|
215 |
afanfani |
1.1 |
########################################################################
|