ViewVC Help
View File | Revision Log | Show Annotations | Root Listing
root/cvsroot/COMP/CRAB/python/DataDiscovery.py
(Generate patch)

Comparing COMP/CRAB/python/DataDiscovery.py (file contents):
Revision 1.13 by slacapra, Mon Dec 18 12:28:06 2006 UTC vs.
Revision 1.21 by spiga, Thu Jun 5 07:13:31 2008 UTC

# Line 1 | Line 1
1   #!/usr/bin/env python
2 < import sys, os, string, re
3 < from DBSInfo import *
2 > import exceptions
3 > import DBSAPI.dbsApi
4 > from DBSAPI.dbsApiException import *
5 > import common
6 > from crab_util import *
7  
8  
9 + # #######################################
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   # ####################################
38   class DataDiscoveryError(exceptions.Exception):
39      def __init__(self, errorMessage):
40 <        args=errorMessage
41 <        exceptions.Exception.__init__(self, args)
40 >        self.args=errorMessage
41 >        exceptions.Exception.__init__(self, self.args)
42          pass
43  
44      def getErrorMessage(self):
# Line 17 | Line 48 | class DataDiscoveryError(exceptions.Exce
48   # ####################################
49   class NotExistingDatasetError(exceptions.Exception):
50      def __init__(self, errorMessage):
51 <        args=errorMessage
52 <        exceptions.Exception.__init__(self, args)
51 >        self.args=errorMessage
52 >        exceptions.Exception.__init__(self, self.args)
53          pass
54  
55      def getErrorMessage(self):
# Line 28 | Line 59 | class NotExistingDatasetError(exceptions
59   # ####################################
60   class NoDataTierinProvenanceError(exceptions.Exception):
61      def __init__(self, errorMessage):
62 <        args=errorMessage
63 <        exceptions.Exception.__init__(self, args)
62 >        self.args=errorMessage
63 >        exceptions.Exception.__init__(self, self.args)
64          pass
65  
66      def getErrorMessage(self):
# Line 39 | Line 70 | class NoDataTierinProvenanceError(except
70   # ####################################
71   # class to find and extact info from published data
72   class DataDiscovery:
73 <    def __init__(self, datasetPath, dataTiers, cfg_params):
73 >    def __init__(self, datasetPath, cfg_params):
74  
75 < #       Attributes
75 >        #       Attributes
76          self.datasetPath = datasetPath
46        self.dataTiers = dataTiers
77          self.cfg_params = cfg_params
78  
79          self.eventsPerBlock = {}  # DBS output: map fileblocks-events for collection
80          self.eventsPerFile = {}   # DBS output: map files-events
81 <        self.blocksinfo = {}  # DBS output: map fileblocks-files
82 < #DBS output: max events computed by method getMaxEvents
81 >        self.blocksinfo = {}      # DBS output: map fileblocks-files
82 >        self.maxEvents = 0        # DBS output: max events
83 >        self.parent = {}       # DBS output: max events
84  
85   # ####################################
86      def fetchDBSInfo(self):
# Line 58 | Line 89 | class DataDiscovery:
89          """
90  
91          ## get DBS URL
92 <        try:
92 >        dbs_url="http://cmsdbsprod.cern.ch/cms_dbs_prod_global/servlet/DBSServlet"
93 >        if (self.cfg_params.has_key('CMSSW.dbs_url')):
94              dbs_url=self.cfg_params['CMSSW.dbs_url']
63        except KeyError:
64            dbs_url="http://cmsdoc.cern.ch/cms/test/aprom/DBS/CGIServer/prodquery"
95  
96 <        ## get info about the requested dataset
97 <        try:
98 <            dbs_instance=self.cfg_params['CMSSW.dbs_instance']
99 <        except KeyError:
100 <            dbs_instance="MCGlobal/Writer"
101 <
102 <        dbs = DBSInfo(dbs_url, dbs_instance)
96 >        common.logger.debug(3,"Accessing DBS at: "+dbs_url)
97 >
98 >        ## check if runs are selected
99 >        runselection = []
100 >        if (self.cfg_params.has_key('CMSSW.runselection')):
101 >            runselection = parseRange2(self.cfg_params['CMSSW.runselection'])
102 >
103 >        common.logger.debug(6,"runselection is: %s"%runselection)
104 >        ## service API
105 >        args = {}
106 >        args['url']     = dbs_url
107 >        args['level']   = 'CRITICAL'
108 >
109 >        ## 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 >        api = DBSAPI.dbsApi.DbsApi(args)
125          try:
126 <            self.datasets = dbs.getMatchingDatasets(self.datasetPath)
127 <        except dbsCgiApi.DbsCgiExecutionError, msg:
126 >            if len(runselection) <= 0 :
127 >                files = api.listFiles(path=self.datasetPath,retriveList=allowedRetriveValue)
128 >            else :
129 >                files=[]
130 >                for arun in runselection:
131 >                    try:
132 >                        filesinrun = api.listFiles(path=self.datasetPath,retriveList=allowedRetriveValue,runNumber=arun)
133 >                        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 >        except DbsBadRequest, msg:
140              raise DataDiscoveryError(msg)
141          except DBSError, msg:
142              raise DataDiscoveryError(msg)
143  
144 <        if len(self.datasets) == 0:
145 <            raise DataDiscoveryError("DatasetPath=%s unknown to DBS" %self.datasetPath)
146 <        if len(self.datasets) > 1:
147 <            raise DataDiscoveryError("DatasetPath=%s is ambiguous" %self.datasetPath)
144 >        # parse files and fill arrays
145 >        for file in files :
146 >            parList = []
147 >            filename = file['LogicalFileName']
148 >            # 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 >            if filename.find('.dat') < 0 :
152 >                fileblock = file['Block']['Name']
153 >                events    = file['NumberOfEvents']
154 >                # 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 >                else :
166 >                    self.blocksinfo[fileblock] = [filename]
167  
168 <        try:
169 <            self.dbsdataset = self.datasets[0].get('datasetPathName')
168 >                # total number of events
169 >                self.maxEvents += events
170 >
171 >        for block in self.eventsPerBlock.keys() :
172 >            common.logger.debug(6,"DBSInfo: total nevts %i in block %s "%(self.eventsPerBlock[block],block))
173  
88            self.eventsPerBlock = dbs.getEventsPerBlock(self.dbsdataset)
89            self.blocksinfo = dbs.getDatasetFileBlocks(self.dbsdataset)
90            self.eventsPerFile = dbs.getEventsPerFile(self.dbsdataset)
91        except DBSError, ex:
92            raise DataDiscoveryError(ex.getErrorMessage())
93        
174          if len(self.eventsPerBlock) <= 0:
175 <            raise NotExistingDatasetError (("\nNo data for %s in DBS\nPlease check"
175 >            raise NotExistingDatasetError(("\nNo data for %s in DBS\nPlease check"
176                                              + " dataset path variables in crab.cfg")
177 <                                            % self.dbsdataset)
177 >                                            % self.datasetPath)
178  
179  
180   # #################################################
# Line 102 | Line 182 | class DataDiscovery:
182          """
183          max events
184          """
185 <        ## loop over the event collections
106 <        nevts=0      
107 <        for evc_evts in self.eventsPerBlock.values():
108 <            nevts=nevts+evc_evts
109 <
110 <        return nevts
185 >        return self.maxEvents
186  
187   # #################################################
188      def getEventsPerBlock(self):
# Line 130 | Line 205 | class DataDiscovery:
205          """
206          return self.blocksinfo        
207  
208 + # #################################################
209 +    def getParent(self):
210 +        """
211 +        return parent grouped by file
212 +        """
213 +        return self.parent        
214 +
215   ########################################################################

Diff Legend

Removed lines
+ Added lines
< Changed lines
> Changed lines