CRAB/python/DataDiscovery.py

#!/usr/bin/env python
from DBSInfo import *


# ####################################
class DataDiscoveryError(exceptions.Exception):
    def __init__(self, errorMessage):
        self.args=errorMessage
        exceptions.Exception.__init__(self, self.args)
        pass

    def getErrorMessage(self):
        """ Return exception error """
        return "%s" % (self.args)

# ####################################
class NotExistingDatasetError(exceptions.Exception):
    def __init__(self, errorMessage):
        self.args=errorMessage
        exceptions.Exception.__init__(self, self.args)
        pass

    def getErrorMessage(self):
        """ Return exception error """
        return "%s" % (self.args)

# ####################################
class NoDataTierinProvenanceError(exceptions.Exception):
    def __init__(self, errorMessage):
        self.args=errorMessage
        exceptions.Exception.__init__(self, self.args)
        pass

    def getErrorMessage(self):
        """ Return exception error """
        return "%s" % (self.args)

# ####################################
# class to find and extact info from published data
class DataDiscovery:
    def __init__(self, datasetPath, cfg_params):

#       Attributes
        self.datasetPath = datasetPath
        self.cfg_params = cfg_params

        self.eventsPerBlock = {}  # DBS output: map fileblocks-events for collection
        self.eventsPerFile = {}   # DBS output: map files-events
        self.blocksinfo = {}  # DBS output: map fileblocks-files 
#DBS output: max events computed by method getMaxEvents 

# ####################################
    def fetchDBSInfo(self):
        """
        Contact DBS
        """

        ## get DBS URL
        try:
            dbs_url=self.cfg_params['CMSSW.dbs_url']
        except KeyError:
            dbs_url="http://cmsdoc.cern.ch/cms/test/aprom/DBS/CGIServer/prodquery"

        ## get info about the requested dataset
        try:
            dbs_instance=self.cfg_params['CMSSW.dbs_instance']
        except KeyError:
            dbs_instance="MCGlobal/Writer"
 
        dbs = DBSInfo(dbs_url, dbs_instance)
        try:
            self.datasets = dbs.getMatchingDatasets(self.datasetPath)
        except DBS1API.dbsCgiApi.DbsCgiExecutionError, msg:
            raise DataDiscoveryError(msg)
        except DBSError, msg:
            raise DataDiscoveryError(msg)

        if len(self.datasets) == 0:
            raise DataDiscoveryError("DatasetPath=%s unknown to DBS" %self.datasetPath)
        if len(self.datasets) > 1:
            raise DataDiscoveryError("DatasetPath=%s is ambiguous" %self.datasetPath)

        try:
            self.dbsdataset = self.datasets[0].get('datasetPathName')

            self.eventsPerBlock = dbs.getEventsPerBlock(self.dbsdataset)
            self.blocksinfo = dbs.getDatasetFileBlocks(self.dbsdataset)
            self.eventsPerFile = dbs.getEventsPerFile(self.dbsdataset)
        except DBSError, ex:
            raise DataDiscoveryError(ex.getErrorMessage())
        
        if len(self.eventsPerBlock) <= 0:
            raise NotExistingDatasetError (("\nNo data for %s in DBS\nPlease check"
                                            + " dataset path variables in crab.cfg")
                                            % self.dbsdataset)


# #################################################
    def getMaxEvents(self):
        """
        max events 
        """
        ## loop over the event collections 
        nevts=0       
        for evc_evts in self.eventsPerBlock.values():
            nevts=nevts+evc_evts

        return nevts

# #################################################
    def getEventsPerBlock(self):
        """
        list the event collections structure by fileblock 
        """
        return self.eventsPerBlock

# #################################################
    def getEventsPerFile(self):
        """
        list the event collections structure by file 
        """
        return self.eventsPerFile

# #################################################
    def getFiles(self):
        """
        return files grouped by fileblock 
        """
        return self.blocksinfo        

########################################################################
Revision:	1.15
Committed:	Thu Feb 1 16:12:40 2007 UTC (18 years, 2 months ago) by gutsche
Content type:	text/x-python
Branch:	MAIN
Changes since 1.14:	+5 -6 lines
Log Message:	Move DBS-1 APIs to DBS1API namespace. This requires preparation of special DBS checkouts: - checkout DBSAPIs - create directory DBSAPI in $CRABDIR - check out COMP/DBS/Clients/PythonAPI into DBSAPI/DBS1API - check out COMP/DBS/Clients/Python into DBSAPI/DBS2API - create namespace __init__.py in both subdirectories DBS1API and DBS2API - change configure to add $CRABDIR/DBSAPI to PYTHONPATH and to set DBS_CLIENT_CONFIG
#	User	Rev	Content
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	gutsche	1.15	self.args=errorMessage
9	slacapra	1.14	exceptions.Exception.__init__(self, self.args)
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	gutsche	1.15	self.args=errorMessage
20	slacapra	1.14	exceptions.Exception.__init__(self, self.args)
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	gutsche	1.15	self.args=errorMessage
31	slacapra	1.14	exceptions.Exception.__init__(self, self.args)
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	gutsche	1.15	def __init__(self, datasetPath, cfg_params):
42	afanfani	1.1
43			# Attributes
44	slacapra	1.11	self.datasetPath = datasetPath
45	afanfani	1.1	self.cfg_params = cfg_params
46
47	slacapra	1.11	self.eventsPerBlock = {} # DBS output: map fileblocks-events for collection
48			self.eventsPerFile = {} # DBS output: map files-events
49			self.blocksinfo = {} # DBS output: map fileblocks-files
50	afanfani	1.1	#DBS output: max events computed by method getMaxEvents
51
52			# ####################################
53			def fetchDBSInfo(self):
54			"""
55			Contact DBS
56			"""
57
58	slacapra	1.11	## get DBS URL
59			try:
60			dbs_url=self.cfg_params['CMSSW.dbs_url']
61			except KeyError:
62			dbs_url="http://cmsdoc.cern.ch/cms/test/aprom/DBS/CGIServer/prodquery"
63	afanfani	1.3
64	slacapra	1.11	## get info about the requested dataset
65			try:
66			dbs_instance=self.cfg_params['CMSSW.dbs_instance']
67			except KeyError:
68	gutsche	1.12	dbs_instance="MCGlobal/Writer"
69	slacapra	1.11
70			dbs = DBSInfo(dbs_url, dbs_instance)
71	afanfani	1.5	try:
72	slacapra	1.11	self.datasets = dbs.getMatchingDatasets(self.datasetPath)
73	gutsche	1.15	except DBS1API.dbsCgiApi.DbsCgiExecutionError, msg:
74	slacapra	1.11	raise DataDiscoveryError(msg)
75	slacapra	1.13	except DBSError, msg:
76			raise DataDiscoveryError(msg)
77	slacapra	1.11
78	afanfani	1.4	if len(self.datasets) == 0:
79	slacapra	1.11	raise DataDiscoveryError("DatasetPath=%s unknown to DBS" %self.datasetPath)
80	afanfani	1.4	if len(self.datasets) > 1:
81	slacapra	1.11	raise DataDiscoveryError("DatasetPath=%s is ambiguous" %self.datasetPath)
82
83	afanfani	1.1	try:
84	slacapra	1.7	self.dbsdataset = self.datasets[0].get('datasetPathName')
85	slacapra	1.11
86			self.eventsPerBlock = dbs.getEventsPerBlock(self.dbsdataset)
87			self.blocksinfo = dbs.getDatasetFileBlocks(self.dbsdataset)
88			self.eventsPerFile = dbs.getEventsPerFile(self.dbsdataset)
89	afanfani	1.3	except DBSError, ex:
90	slacapra	1.7	raise DataDiscoveryError(ex.getErrorMessage())
91	afanfani	1.3
92	slacapra	1.11	if len(self.eventsPerBlock) <= 0:
93			raise NotExistingDatasetError (("\nNo data for %s in DBS\nPlease check"
94			+ " dataset path variables in crab.cfg")
95			% self.dbsdataset)
96	afanfani	1.1
97
98			# #################################################
99			def getMaxEvents(self):
100			"""
101	slacapra	1.11	max events
102	afanfani	1.1	"""
103	slacapra	1.11	## loop over the event collections
104	afanfani	1.1	nevts=0
105	slacapra	1.11	for evc_evts in self.eventsPerBlock.values():
106			nevts=nevts+evc_evts
107	afanfani	1.1
108			return nevts
109
110			# #################################################
111	slacapra	1.11	def getEventsPerBlock(self):
112	afanfani	1.1	"""
113	slacapra	1.11	list the event collections structure by fileblock
114	afanfani	1.1	"""
115	slacapra	1.11	return self.eventsPerBlock
116	afanfani	1.1
117			# #################################################
118	slacapra	1.11	def getEventsPerFile(self):
119	afanfani	1.1	"""
120	slacapra	1.11	list the event collections structure by file
121	afanfani	1.1	"""
122	slacapra	1.11	return self.eventsPerFile
123	afanfani	1.1
124			# #################################################
125	slacapra	1.11	def getFiles(self):
126	afanfani	1.1	"""
127	slacapra	1.11	return files grouped by fileblock
128	afanfani	1.1	"""
129	slacapra	1.11	return self.blocksinfo
130	afanfani	1.1
131			########################################################################