99 |
|
'glidein': global_url,\ |
100 |
|
'lsf': global_url,\ |
101 |
|
'caf': caf_url,\ |
102 |
< |
'sge': global_url |
102 |
> |
'sge': global_url, |
103 |
> |
'arc': global_url |
104 |
|
} |
105 |
|
|
106 |
|
dbs_url_default = dbs_url_map[(common.scheduler.name()).lower()] |
112 |
|
if (self.cfg_params.has_key('CMSSW.runselection')): |
113 |
|
runselection = parseRange2(self.cfg_params['CMSSW.runselection']) |
114 |
|
|
115 |
+ |
|
116 |
+ |
self.splitByRun = int(self.cfg_params.get('CMSSW.split_by_run', 0)) |
117 |
+ |
|
118 |
+ |
self.ads = int(self.cfg_params.get('CMSSW.ads', 0)) |
119 |
+ |
|
120 |
|
common.logger.debug(6,"runselection is: %s"%runselection) |
121 |
|
## service API |
122 |
|
args = {} |
124 |
|
args['level'] = 'CRITICAL' |
125 |
|
|
126 |
|
## check if has been requested to use the parent info |
127 |
< |
useParent = self.cfg_params.get('CMSSW.use_parent',False) |
127 |
> |
useparent = int(self.cfg_params.get('CMSSW.use_parent',0)) |
128 |
|
|
129 |
|
## check if has been asked for a non default file to store/read analyzed fileBlocks |
130 |
|
defaultName = common.work_space.shareDir()+'AnalyzedBlocks.txt' |
131 |
|
fileBlocks_FileName = os.path.abspath(self.cfg_params.get('CMSSW.fileblocks_file',defaultName)) |
132 |
|
|
133 |
|
api = DBSAPI.dbsApi.DbsApi(args) |
128 |
– |
allowedRetriveValue = ['retrive_parent', |
129 |
– |
'retrive_block', |
130 |
– |
'retrive_lumi', |
131 |
– |
'retrive_run' |
132 |
– |
] |
133 |
– |
try: |
134 |
– |
if len(runselection) <= 0 : |
135 |
– |
if useParent: |
136 |
– |
files = api.listFiles(path=self.datasetPath, retriveList=allowedRetriveValue) |
137 |
– |
common.logger.debug(5,"Set of input parameters used for DBS query : \n"+str(allowedRetriveValue)) |
138 |
– |
common.logger.write("Set of input parameters used for DBS query : \n"+str(allowedRetriveValue)) |
139 |
– |
else: |
140 |
– |
files = api.listDatasetFiles(self.datasetPath) |
141 |
– |
else : |
142 |
– |
files=[] |
143 |
– |
for arun in runselection: |
144 |
– |
try: |
145 |
– |
filesinrun = api.listFiles(path=self.datasetPath,retriveList=allowedRetriveValue,runNumber=arun) |
146 |
– |
files.extend(filesinrun) |
147 |
– |
except: |
148 |
– |
msg="WARNING: problem extracting info from DBS for run %s "%arun |
149 |
– |
common.logger.message(msg) |
150 |
– |
pass |
134 |
|
|
135 |
< |
except DbsBadRequest, msg: |
153 |
< |
raise DataDiscoveryError(msg) |
154 |
< |
except DBSError, msg: |
155 |
< |
raise DataDiscoveryError(msg) |
135 |
> |
self.files = self.queryDbs(api,path=self.datasetPath,runselection=runselection,useParent=useparent) |
136 |
|
|
137 |
|
anFileBlocks = [] |
138 |
|
if self.skipBlocks: anFileBlocks = readTXTfile(self, fileBlocks_FileName) |
139 |
|
|
140 |
|
# parse files and fill arrays |
141 |
< |
for file in files : |
141 |
> |
for file in self.files : |
142 |
|
parList = [] |
143 |
|
# skip already analyzed blocks |
144 |
|
fileblock = file['Block']['Name'] |
145 |
|
if fileblock not in anFileBlocks : |
146 |
|
filename = file['LogicalFileName'] |
147 |
|
# asked retry the list of parent for the given child |
148 |
< |
if useParent: parList = [x['LogicalFileName'] for x in file['ParentList']] |
148 |
> |
if useparent==1: parList = [x['LogicalFileName'] for x in file['ParentList']] |
149 |
|
self.parent[filename] = parList |
150 |
|
if filename.find('.dat') < 0 : |
151 |
|
events = file['NumberOfEvents'] |
181 |
|
% self.datasetPath) |
182 |
|
|
183 |
|
|
184 |
+ |
########################### |
185 |
+ |
|
186 |
+ |
def queryDbs(self,api,path=None,runselection=None,useParent=None): |
187 |
+ |
|
188 |
+ |
allowedRetriveValue = [#'retrive_parent', |
189 |
+ |
'retrive_block', |
190 |
+ |
#'retrive_lumi', |
191 |
+ |
'retrive_run' |
192 |
+ |
] |
193 |
+ |
try: |
194 |
+ |
if len(runselection) <=0 : |
195 |
+ |
if useParent==1 or self.splitByRun==1 : |
196 |
+ |
if self.ads==1 : |
197 |
+ |
files = api.listFiles(analysisDataset=path, retriveList=allowedRetriveValue) |
198 |
+ |
else : |
199 |
+ |
files = api.listFiles(path=path, retriveList=allowedRetriveValue) |
200 |
+ |
common.logger.debug(5,"Set of input parameters used for DBS query : \n"+str(allowedRetriveValue)) |
201 |
+ |
common.logger.write("Set of input parameters used for DBS query : \n"+str(allowedRetriveValue)) |
202 |
+ |
else: |
203 |
+ |
files = api.listDatasetFiles(self.datasetPath) |
204 |
+ |
else : |
205 |
+ |
files=[] |
206 |
+ |
for arun in runselection: |
207 |
+ |
try: |
208 |
+ |
if self.ads==1 : filesinrun = api.listFiles(analysisDataset=path,retriveList=allowedRetriveValue,runNumber=arun) |
209 |
+ |
else: filesinrun = api.listFiles(path=path,retriveList=allowedRetriveValue,runNumber=arun) |
210 |
+ |
files.extend(filesinrun) |
211 |
+ |
except: |
212 |
+ |
msg="WARNING: problem extracting info from DBS for run %s "%arun |
213 |
+ |
common.logger.message(msg) |
214 |
+ |
pass |
215 |
+ |
|
216 |
+ |
except DbsBadRequest, msg: |
217 |
+ |
raise DataDiscoveryError(msg) |
218 |
+ |
except DBSError, msg: |
219 |
+ |
raise DataDiscoveryError(msg) |
220 |
+ |
|
221 |
+ |
return files |
222 |
+ |
|
223 |
|
# ################################################# |
224 |
|
def getMaxEvents(self): |
225 |
|
""" |
255 |
|
""" |
256 |
|
return self.parent |
257 |
|
|
258 |
+ |
# ################################################# |
259 |
+ |
def getListFiles(self): |
260 |
+ |
""" |
261 |
+ |
return parent grouped by file |
262 |
+ |
""" |
263 |
+ |
return self.files |
264 |
+ |
|
265 |
|
######################################################################## |