Python Lexicon.block方法代码示例

# 需要导入模块: from WMCore import Lexicon [as 别名]
# 或者: from WMCore.Lexicon import block [as 别名]
    def validBlocks(self, task, dbs):
        """Return blocks that pass the input data restriction"""
        datasetPath = task.getInputDatasetPath()
        validBlocks = []

        blockWhiteList = task.inputBlockWhitelist()
        blockBlackList = task.inputBlockBlacklist()
        runWhiteList = task.inputRunWhitelist()
        runBlackList = task.inputRunBlacklist()
        if task.getLumiMask():  # if we have a lumi mask get only the relevant blocks
            maskedBlocks = self.getMaskedBlocks(task, dbs, datasetPath)
        if task.getTrustSitelists().get('trustlists'):
            siteWhitelist = task.siteWhitelist()
            siteBlacklist = task.siteBlacklist()
            self.sites = makeLocationsList(siteWhitelist, siteBlacklist)

        blocks = []
        # Take data inputs or from spec
        if not self.data:
            if blockWhiteList:
                self.data = dict((block, []) for block in blockWhiteList)
            else:
                self.data = {datasetPath: []}  # same structure as in WorkQueueElement

        for data in self.data:
            if data.find('#') > -1:
                Lexicon.block(data)  # check block name
                datasetPath = str(data.split('#')[0])
                blocks.append(str(data))
            else:
                Lexicon.dataset(data)  # check dataset name
                for block in dbs.listFileBlocks(data, onlyClosedBlocks=True):
                    blocks.append(str(block))

        for blockName in blocks:
            # check block restrictions
            if blockWhiteList and blockName not in blockWhiteList:
                continue
            if blockName in blockBlackList:
                continue
            if blockName in self.blockBlackListModifier:
                # Don't duplicate blocks rejected before or blocks that were included and therefore are now in the blacklist
                continue
            if task.getLumiMask() and blockName not in maskedBlocks:
                self.rejectedWork.append(blockName)
                continue

            block = dbs.getDBSSummaryInfo(datasetPath, block=blockName)
            # blocks with 0 valid files should be ignored
            # - ideally they would be deleted but dbs can't delete blocks
            if not block['NumberOfFiles'] or block['NumberOfFiles'] == '0':
                self.rejectedWork.append(blockName)
                continue

            # check lumi restrictions
            if task.getLumiMask():
                accepted_lumis = sum([len(maskedBlocks[blockName][lfn].getLumis()) for lfn in maskedBlocks[blockName]])
                # use the information given from getMaskedBlocks to compute che size of the block
                block['NumberOfFiles'] = len(maskedBlocks[blockName])
                # ratio =  lumis which are ok in the block / total num lumis
                ratioAccepted = accepted_lumis / block['NumberOfLumis']
                block['NumberOfEvents'] = block['NumberOfEvents'] * ratioAccepted
                block[self.lumiType] = accepted_lumis
            # check run restrictions
            elif runWhiteList or runBlackList:
                # listRunLumis returns a dictionary with the lumi sections per run
                runLumis = dbs.listRunLumis(block=block['block'])
                runs = set(runLumis.keys())
                recalculateLumiCounts = False
                if len(runs) > 1:
                    # If more than one run in the block
                    # Then we must calculate the lumi counts after filtering the run list
                    # This has to be done rarely and requires calling DBS file information
                    recalculateLumiCounts = True

                # apply blacklist
                runs = runs.difference(runBlackList)
                # if whitelist only accept listed runs
                if runWhiteList:
                    runs = runs.intersection(runWhiteList)
                # any runs left are ones we will run on, if none ignore block
                if not runs:
                    self.rejectedWork.append(blockName)
                    continue

                if len(runs) == len(runLumis):
                    # If there is no change in the runs, then we can skip recalculating lumi counts
                    recalculateLumiCounts = False

                if recalculateLumiCounts:
                    # Recalculate effective size of block
                    # We pull out file info, since we don't do this often
                    acceptedLumiCount = 0
                    acceptedEventCount = 0
                    acceptedFileCount = 0
                    fileInfo = dbs.listFilesInBlock(fileBlockName=block['block'])
                    for fileEntry in fileInfo:
                        acceptedFile = False
                        acceptedFileLumiCount = 0
                        for lumiInfo in fileEntry['LumiList']:
#.........这里部分代码省略.........

# 需要导入模块: from WMCore import Lexicon [as 别名]
# 或者: from WMCore.Lexicon import block [as 别名]
    def validBlocks(self, task, dbs):
        """Return blocks that pass the input data restriction"""
        datasetPath = task.getInputDatasetPath()
        validBlocks = []

        blockWhiteList = task.inputBlockWhitelist()
        blockBlackList = task.inputBlockBlacklist()
        runWhiteList = task.inputRunWhitelist()
        runBlackList = task.inputRunBlacklist()
        if task.getLumiMask(): #if we have a lumi mask get only the relevant blocks
            maskedBlocks = self.getMaskedBlocks(task, dbs, datasetPath)

        blocks = []
        # Take data inputs or from spec
        if not self.data:
            if blockWhiteList:
                self.data = dict((block, []) for block in blockWhiteList)
            else:
                self.data = {datasetPath : []} # same structure as in WorkQueueElement

        for data in self.data:
            if data.find('#') > -1:
                Lexicon.block(data) # check block name
                datasetPath = str(data.split('#')[0])
                blocks.append(str(data))
            else:
                Lexicon.dataset(data) # check dataset name
                for block in dbs.listFileBlocks(data):
                    blocks.append(str(block))


        for blockName in blocks:
            # check block restrictions
            if blockWhiteList and blockName not in blockWhiteList:
                continue
            if blockName in blockBlackList:
                continue
            if task.getLumiMask() and blockName not in maskedBlocks:
                continue

            block = dbs.getDBSSummaryInfo(datasetPath, block = blockName)
            # blocks with 0 valid files should be ignored
            # - ideally they would be deleted but dbs can't delete blocks
            if not block['NumberOfFiles'] or block['NumberOfFiles'] == '0':
                continue

            #check lumi restrictions
            if task.getLumiMask():
                accepted_lumis = sum( [ len(maskedBlocks[blockName][file]) for file in maskedBlocks[blockName] ] )
                #use the information given from getMaskedBlocks to compute che size of the block
                block['NumberOfFiles'] = len(maskedBlocks[blockName])
                #ratio =  lumis which are ok in the block / total num lumis
                ratio_accepted = 1. * accepted_lumis / float(block['NumberOfLumis'])
                block['NumberOfEvents'] = float(block['NumberOfEvents']) * ratio_accepted
                block[self.lumiType] = accepted_lumis
            # check run restrictions
            elif runWhiteList or runBlackList:
                # listRuns returns a run number per lumi section
                full_lumi_list = dbs.listRuns(block = block['block'])
                runs = set(full_lumi_list)

                # apply blacklist
                runs = runs.difference(runBlackList)
                # if whitelist only accept listed runs
                if runWhiteList:
                    runs = runs.intersection(runWhiteList)

                # any runs left are ones we will run on, if none ignore block
                if not runs:
                    continue

                # recalculate effective size of block
                # make a guess for new event/file numbers from ratio
                # of accepted lumi sections (otherwise have to pull file info)
                accepted_lumis = [x for x in full_lumi_list if x in runs]
                ratio_accepted = 1. * len(accepted_lumis) / len(full_lumi_list)
                block[self.lumiType] = len(accepted_lumis)
                block['NumberOfFiles'] = float(block['NumberOfFiles']) * ratio_accepted
                block['NumberOfEvents'] = float(block['NumberOfEvents']) * ratio_accepted

            # save locations
            self.data[block['block']] = sitesFromStorageEelements(dbs.listFileBlockLocation(block['block']))

            validBlocks.append(block)
        return validBlocks

# 需要导入模块: from WMCore import Lexicon [as 别名]
# 或者: from WMCore.Lexicon import block [as 别名]
    def validBlocks(self, task, dbs):
        """Return blocks that pass the input data restriction"""
        datasetPath = task.getInputDatasetPath()
        validBlocks = []

        blockWhiteList = task.inputBlockWhitelist()
        blockBlackList = task.inputBlockBlacklist()
        runWhiteList = task.inputRunWhitelist()
        runBlackList = task.inputRunBlacklist()
        if task.getLumiMask(): #if we have a lumi mask get only the relevant blocks
            maskedBlocks = self.getMaskedBlocks(task, dbs, datasetPath)
        if task.inputLocationFlag():
            # Then get the locations from the site whitelist/blacklist + SiteDB
            siteWhitelist = task.siteWhitelist()
            siteBlacklist = task.siteBlacklist()
            if siteWhitelist:
                # Just get the ses matching the whitelists
                self.sites = siteWhitelist
            elif siteBlacklist:
                # Get all CMS sites less the blacklist
                allSites = cmsSiteNames()
                self.sites = list(set(allSites) - set (siteBlacklist))
            else:
                # Run at any CMS site
                self.sites = cmsSiteNames()

        blocks = []
        # Take data inputs or from spec
        if not self.data:
            if blockWhiteList:
                self.data = dict((block, []) for block in blockWhiteList)
            else:
                self.data = {datasetPath : []} # same structure as in WorkQueueElement

        for data in self.data:
            if data.find('#') > -1:
                Lexicon.block(data) # check block name
                datasetPath = str(data.split('#')[0])
                blocks.append(str(data))
            else:
                Lexicon.dataset(data) # check dataset name
                for block in dbs.listFileBlocks(data):
                    blocks.append(str(block))


        for blockName in blocks:
            # check block restrictions
            if blockWhiteList and blockName not in blockWhiteList:
                continue
            if blockName in blockBlackList:
                continue
            if task.getLumiMask() and blockName not in maskedBlocks:
                continue

            block = dbs.getDBSSummaryInfo(datasetPath, block = blockName)
            # blocks with 0 valid files should be ignored
            # - ideally they would be deleted but dbs can't delete blocks
            if not block['NumberOfFiles'] or block['NumberOfFiles'] == '0':
                continue

            #check lumi restrictions
            if task.getLumiMask():
                accepted_lumis = sum( [ len(maskedBlocks[blockName][file]) for file in maskedBlocks[blockName] ] )
                #use the information given from getMaskedBlocks to compute che size of the block
                block['NumberOfFiles'] = len(maskedBlocks[blockName])
                #ratio =  lumis which are ok in the block / total num lumis
                ratioAccepted = 1. * accepted_lumis / float(block['NumberOfLumis'])
                block['NumberOfEvents'] = float(block['NumberOfEvents']) * ratioAccepted
                block[self.lumiType] = accepted_lumis
            # check run restrictions
            elif runWhiteList or runBlackList:
                # listRunLumis returns a dictionary with the lumi sections per run
                runLumis = dbs.listRunLumis(block = block['block'])
                runs = set(runLumis.keys())
                recalculateLumiCounts = False
                if len(runs) > 1:
                    # If more than one run in the block
                    # Then we must calculate the lumi counts after filtering the run list
                    # This has to be done rarely and requires calling DBS file information
                    recalculateLumiCounts = True

                # apply blacklist
                runs = runs.difference(runBlackList)
                # if whitelist only accept listed runs
                if runWhiteList:
                    runs = runs.intersection(runWhiteList)
                # any runs left are ones we will run on, if none ignore block
                if not runs:
                    continue

                if len(runs) == len(runLumis):
                    # If there is no change in the runs, then we can skip recalculating lumi counts
                    recalculateLumiCounts = False

                if recalculateLumiCounts:
                    # Recalculate effective size of block
                    # We pull out file info, since we don't do this often
                    acceptedLumiCount = 0
                    acceptedEventCount = 0
                    acceptedFileCount = 0
#.........这里部分代码省略.........

# 需要导入模块: from WMCore import Lexicon [as 别名]
# 或者: from WMCore.Lexicon import block [as 别名]
    def validBlocks(self, task, dbs):
        """Return blocks that pass the input data restriction"""
        datasetPath = task.getInputDatasetPath()
        validBlocks = []

        blockWhiteList = task.inputBlockWhitelist()
        blockBlackList = task.inputBlockBlacklist()
        runWhiteList = task.inputRunWhitelist()
        runBlackList = task.inputRunBlacklist()

        blocks = []
        # Take data inputs or from spec
        if not self.data:
            self.data = {datasetPath : []} # same structure as in WorkQueueElement
            #blocks = dbs.getFileBlocksInfo(datasetPath, locations = False)
        #else:
            #dataItems = self.data.keys()

        for data in self.data:
            if data.find('#') > -1:
                Lexicon.block(data) # check dataset name
                datasetPath = str(data.split('#')[0])
                blocks.extend(dbs.getFileBlocksInfo(datasetPath, blockName = str(data), locations = True))
            else:
                Lexicon.dataset(data) # check dataset name
                blocks.extend(dbs.getFileBlocksInfo(datasetPath, locations = True))

        for block in blocks:
            # blocks with 0 valid files should be ignored
            # - ideally they would be deleted but dbs can't delete blocks
            if not block['NumberOfFiles']:
                continue

            # check block restrictions
            if blockWhiteList and block['Name'] not in blockWhiteList:
                continue
            if block['Name'] in blockBlackList:
                continue

            # check run restrictions
            if runWhiteList or runBlackList:
                # listRuns returns a run number per lumi section
                full_lumi_list = dbs.listRuns(block = block['Name'])
                runs = set(full_lumi_list)

                # apply blacklist
                runs = runs.difference(runBlackList)
                # if whitelist only accept listed runs
                if runWhiteList:
                    runs = runs.intersection(runWhiteList)

                # any runs left are ones we will run on, if none ignore block
                if not runs:
                    continue

                # recalculate effective size of block
                # make a guess for new event/file numbers from ratio
                # of accepted lumi sections (otherwise have to pull file info)
                accepted_lumis = [x for x in full_lumi_list if x in runs]
                ratio_accepted = 1. * len(accepted_lumis) / len(full_lumi_list)
                block[self.lumiType] = len(accepted_lumis)
                block['NumberOfFiles'] *= ratio_accepted
                block['NumberOfEvents'] *= ratio_accepted

            # get lumi info if needed and not already available
            if self.args['SliceType'] == self.lumiType and not block.get(self.lumiType):
                blockSummary = dbs.getDBSSummaryInfo(block = block["Name"])
                block[self.lumiType] = blockSummary[self.lumiType]

            # save locations
            self.data[block['Name']] = sitesFromStorageEelements([x['Name'] for x in block['StorageElementList']])

            validBlocks.append(block)
        return validBlocks