UserCode/RateMonShiftTool_dev/MenuAnalyzer.py

#!/usr/bin/env python

import os
import sys
from DatabaseParser import ConnectDB
import re
import cx_Oracle

try:  ## set is builtin in python 2.6.4 and sets is deprecated
    set
except NameError:
    from sets import Set

class MenuAnalyzer:
    def __init__(self,name):
        ##default ranges
        self.maxModuleNameLength = 300
        self.maxModulesPerPath   = 50
        self.maxPaths = 500
        self.maxEndPaths = 30

        ##required streams
        self.requiredStreamsAndPDs = { 'Calibration' : ['TestEnablesEcalHcalDT'],'EcalCalibration' : ['EcalLaser'],
                                  'TrackerCalibration' : ['TestEnablesTracker'], 'DQM' : 'OnlineMonitor'}
        self.ExpressStreamName = 'Express'
        self.expressPDs      = { 'ExpressPhysics' : 'Collisions',
                                 'ExpressCosmics' : 'Cosmics' }

        self.expressType = ''

        self.menuName = name
        self.endPathList = set()
        self.perPathModuleList={}
        self.perModuleTypeList={}
        self.perStreamPDList={}
        self.perPDPathList={}
        self.Results={}
        self.ModuleList=[]
        self.AnalysisList=[]
        

        ## statically define the analysis map: new analyses must be registered here
        self.AnalysisMap = {
            'moduleLength' : self.checkModuleLength,
            'numberOfPaths' : self.checkNumPaths,
            'numberOfEndPaths' : self.checkNumEndPaths,
            'reqStreamsAndPDs' : self.reqStreamsAndPDs,
            'checkExpress' : self.checkExpress,
            'checkNameFormats' :self.checkNameFormats
            }
        self.ProblemDescriptions = {
            'moduleLength':'Modules too long',
            'numberOfPaths':'Too many paths',
            'numberOfEndPaths':'Too many endpaths',
            'reqStreamsAndPDs':'Missing required stream/PD',
            'checkExpress' : 'Invalid or missing express stream/PD',
            'checkNameFormats' : 'Invalid PD or path name format'
            }

        self.T0REGEXP = { ## These are the regexps that T0 uses to access things
            # Regexp for save file name paths. We don't process anything else.
            'RXSAFEPATH' : re.compile("^[-A-Za-z0-9_]+$"),
            # Regexp for valid dataset names.
            'RXDATASET' : re.compile("^[-A-Za-z0-9_]+$"),
            # Regexp for valid RelVal dataset names.
            'RXRELVALMC' : re.compile("^/RelVal[^/]+/(CMSSW(?:_[0-9]+)+(?:_pre[0-9]+)?)[-_].*$"),
            'RXRELVALDATA' : re.compile("^/[^/]+/(CMSSW(?:_[0-9]+)+(?:_pre[0-9]+)?)[-_].*$"),
            # Regexp for online DQM files.
            'RXONLINE' : re.compile("^(?:.*/)?DQM_V(\d+)(_[A-Za-z0-9]+)?_R(\d+)\.root$"),
            # Regexp for offline DQM files.
            'RXOFFLINE' : re.compile("^(?:.*/)?DQM_V(\d+)_R(\d+)((?:__[-A-Za-z0-9_]+){3})\.root$"),
            # Regexp for acquisition era part of the processed dataset name.
            'RXERA' : re.compile("^([A-Za-z]+\d+|CMSSW(?:_[0-9]+)+(?:_pre[0-9]+)?)")
            }

    def AddAnalysis(self,name): 
        self.AnalysisList.append(name)

    def AddAllAnalyses(self):
        for name in self.AnalysisMap.iterkeys():
            self.AddAnalysis(name)

    def Analyze(self):
        cursor = ConnectDB('hlt')
        self.GetModules(cursor)
        self.GetStreamsPathsPDs(cursor)
        for analysis in self.AnalysisList: 
            if not self.AnalysisMap.has_key(analysis):
                print "ERROR: Analysis %s not defined" % (analysis,)
                continue
            self.AnalysisMap[analysis]()
            
        
    def checkModuleLength(self):
        self.Results['moduleLength'] = []
        for modName,type in self.perModuleTypeList.iteritems():
            if len(modName) > self.maxModuleNameLength: self.Results['moduleLength'].append(modName)
        
    def checkNumPaths(self):
        if len(self.perPathModuleList) > self.maxPaths:
            self.Results['numberOfPaths'] = len(self.perPathModuleList) 
        else:
            self.Results['numberOfPaths'] = 0
    def checkNumEndPaths(self):
        if len(self.endPathList) > self.maxEndPaths:
            self.Results['numberOfEndPaths'] = len(self.endPathList)
        else:
            self.Results['numberOfEndPaths'] = 0
    def reqStreamsAndPDs(self):
        self.Results['reqStreamsAndPDs'] = []
        for stream,PDList in self.requiredStreamsAndPDs.iteritems():
            if not self.perStreamPDList.has_key(stream): self.Results['reqStreamsAndPDs'].append(stream)
            for PD in PDList:
                if not PD in self.requiredStreamsAndPDs[stream]: self.Results['reqStreamsAndPDs'].append(stream+'::'+PD)
    
    def checkExpress(self):
        self.Results['checkExpress'] = []
        if not self.perStreamPDList.has_key(self.ExpressStreamName):
            self.Results['checkExpress'].append(self.ExpressStreamName)
            return

        if len(self.perStreamPDList[self.ExpressStreamName]) >1:
            self.Results['checkExpress'].append("MULTIPLE_PDS")
        if len(self.perStreamPDList[self.ExpressStreamName]) <1:
            self.Results['checkExpress'].append("NO_PDS")

        for PD in self.perStreamPDList[self.ExpressStreamName]:
            if not self.expressPDs.has_key(PD):
                self.Results['checkExpress'].append(self.ExpressStreamName+"::"+PD)
            else:
                self.expressType = self.expressPDs[PD]
                
        
    def checkNameFormats(self):
        self.Results['checkNameFormats']=[]
        for PD in self.perPDPathList.iterkeys():
            if not self.T0REGEXP['RXDATASET'].match(PD):
                self.Results['checkNameFormats'].append(PD)
        for path in self.perPathModuleList.iterkeys():
            if not self.T0REGEXP['RXSAFEPATH'].match(path):
                self.Results['checkNameFormats'].append(path)
        
    def GetModules(self,cursor):
        sqlquery ="""  
        SELECT I.NAME,E.NAME,D.NAME,I.ISENDPATH
        FROM
        CMS_HLT.PARAMETERS B,
        CMS_HLT.SUPERIDPARAMETERASSOC C,
        CMS_HLT.MODULETEMPLATES D,
        CMS_HLT.MODULES E,
        CMS_HLT.PATHMODULEASSOC F,
        CMS_HLT.CONFIGURATIONPATHASSOC G,
        CMS_HLT.CONFIGURATIONS H,
        CMS_HLT.PATHS I
        WHERE
        B.PARAMID = C.PARAMID AND
        C.SUPERID = F.MODULEID AND
        E.TEMPLATEID = D.SUPERID AND
        F.MODULEID = E.SUPERID AND
        F.PATHID=G.PATHID AND
        I.PATHID=G.PATHID AND
        G.CONFIGID=H.CONFIGID AND
        H.CONFIGDESCRIPTOR='%s' 
        """ % (self.menuName,)
        
        cursor.execute(sqlquery)
        for PathName,ModuleName,ModuleType,endPath in cursor.fetchall():
            if not self.perPathModuleList.has_key(PathName): self.perPathModuleList[PathName] = []
            self.perPathModuleList[PathName].append(ModuleName)
            self.perModuleTypeList[ModuleName] = ModuleType
            if endPath: self.endPathList.add(PathName)


    def GetStreamsPathsPDs(self,cursor):
        sqlquery= """
        SELECT A.STREAMLABEL,E.NAME,F.DATASETLABEL
        FROM
        CMS_HLT.STREAMS A,
        CMS_HLT.CONFIGURATIONS B,
        CMS_HLT.CONFIGURATIONPATHASSOC C,
        CMS_HLT.PATHSTREAMDATASETASSOC D,
        CMS_HLT.PATHS E,
        CMS_HLT.PRIMARYDATASETS F
        WHERE
        B.CONFIGDESCRIPTOR='%s' AND
        C.CONFIGID=B.CONFIGID AND
        D.PATHID=C.PATHID AND
        A.STREAMID=D.STREAMID AND
        E.PATHID = C.PATHID AND
        F.DATASETID = D.DATASETID
        """ % (self.menuName,)
        
        cursor.execute(sqlquery)
        for StreamName,PathName,PDName in cursor.fetchall():
            if not self.perStreamPDList.has_key(StreamName): self.perStreamPDList[StreamName] = []
            if not PDName in self.perStreamPDList[StreamName]: self.perStreamPDList[StreamName].append(PDName)
            if not self.perPDPathList.has_key(PDName): self.perPDPathList[PDName] = []
            self.perPDPathList[PDName].append(PathName)

Revision:	1.3
Committed:	Thu Jul 12 16:41:14 2012 UTC (12 years, 9 months ago) by amott
Content type:	text/x-python
Branch:	MAIN
CVS Tags:	MenuAnalyzer_V00-00-01, MenuAnalyzer_V1
Changes since 1.2:	+1 -1 lines
Log Message:	added DQM to required streams
#	User	Rev	Content
1	amott	1.1	#!/usr/bin/env python
2
3			import os
4			import sys
5			from DatabaseParser import ConnectDB
6			import re
7			import cx_Oracle
8
9			try: ## set is builtin in python 2.6.4 and sets is deprecated
10			set
11			except NameError:
12			from sets import Set
13
14			class MenuAnalyzer:
15			def __init__(self,name):
16			##default ranges
17			self.maxModuleNameLength = 300
18			self.maxModulesPerPath = 50
19			self.maxPaths = 500
20			self.maxEndPaths = 30
21
22			##required streams
23			self.requiredStreamsAndPDs = { 'Calibration' : ['TestEnablesEcalHcalDT'],'EcalCalibration' : ['EcalLaser'],
24	amott	1.3	'TrackerCalibration' : ['TestEnablesTracker'], 'DQM' : 'OnlineMonitor'}
25	amott	1.1	self.ExpressStreamName = 'Express'
26			self.expressPDs = { 'ExpressPhysics' : 'Collisions',
27			'ExpressCosmics' : 'Cosmics' }
28
29			self.expressType = ''
30
31			self.menuName = name
32			self.endPathList = set()
33			self.perPathModuleList={}
34			self.perModuleTypeList={}
35			self.perStreamPDList={}
36			self.perPDPathList={}
37			self.Results={}
38			self.ModuleList=[]
39			self.AnalysisList=[]
40
41
42			## statically define the analysis map: new analyses must be registered here
43			self.AnalysisMap = {
44			'moduleLength' : self.checkModuleLength,
45			'numberOfPaths' : self.checkNumPaths,
46			'numberOfEndPaths' : self.checkNumEndPaths,
47			'reqStreamsAndPDs' : self.reqStreamsAndPDs,
48			'checkExpress' : self.checkExpress,
49			'checkNameFormats' :self.checkNameFormats
50			}
51	amott	1.2	self.ProblemDescriptions = {
52			'moduleLength':'Modules too long',
53			'numberOfPaths':'Too many paths',
54			'numberOfEndPaths':'Too many endpaths',
55			'reqStreamsAndPDs':'Missing required stream/PD',
56			'checkExpress' : 'Invalid or missing express stream/PD',
57			'checkNameFormats' : 'Invalid PD or path name format'
58			}
59	amott	1.1
60			self.T0REGEXP = { ## These are the regexps that T0 uses to access things
61			# Regexp for save file name paths. We don't process anything else.
62			'RXSAFEPATH' : re.compile("^[-A-Za-z0-9_]+$"),
63			# Regexp for valid dataset names.
64			'RXDATASET' : re.compile("^[-A-Za-z0-9_]+$"),
65			# Regexp for valid RelVal dataset names.
66			'RXRELVALMC' : re.compile("^/RelVal[^/]+/(CMSSW(?:_[0-9]+)+(?:_pre[0-9]+)?)[-_].*$"),
67			'RXRELVALDATA' : re.compile("^/[^/]+/(CMSSW(?:_[0-9]+)+(?:_pre[0-9]+)?)[-_].*$"),
68			# Regexp for online DQM files.
69			'RXONLINE' : re.compile("^(?:.*/)?DQM_V(\d+)(_[A-Za-z0-9]+)?_R(\d+)\.root$"),
70			# Regexp for offline DQM files.
71			'RXOFFLINE' : re.compile("^(?:.*/)?DQM_V(\d+)_R(\d+)((?:__[-A-Za-z0-9_]+){3})\.root$"),
72			# Regexp for acquisition era part of the processed dataset name.
73			'RXERA' : re.compile("^([A-Za-z]+\d+\|CMSSW(?:_[0-9]+)+(?:_pre[0-9]+)?)")
74			}
75
76			def AddAnalysis(self,name):
77			self.AnalysisList.append(name)
78
79			def AddAllAnalyses(self):
80			for name in self.AnalysisMap.iterkeys():
81			self.AddAnalysis(name)
82
83			def Analyze(self):
84			cursor = ConnectDB('hlt')
85			self.GetModules(cursor)
86			self.GetStreamsPathsPDs(cursor)
87			for analysis in self.AnalysisList:
88			if not self.AnalysisMap.has_key(analysis):
89			print "ERROR: Analysis %s not defined" % (analysis,)
90			continue
91			self.AnalysisMap[analysis]()
92
93
94
95			def checkModuleLength(self):
96			self.Results['moduleLength'] = []
97			for modName,type in self.perModuleTypeList.iteritems():
98			if len(modName) > self.maxModuleNameLength: self.Results['moduleLength'].append(modName)
99
100			def checkNumPaths(self):
101			if len(self.perPathModuleList) > self.maxPaths:
102			self.Results['numberOfPaths'] = len(self.perPathModuleList)
103			else:
104			self.Results['numberOfPaths'] = 0
105			def checkNumEndPaths(self):
106			if len(self.endPathList) > self.maxEndPaths:
107			self.Results['numberOfEndPaths'] = len(self.endPathList)
108			else:
109			self.Results['numberOfEndPaths'] = 0
110			def reqStreamsAndPDs(self):
111			self.Results['reqStreamsAndPDs'] = []
112			for stream,PDList in self.requiredStreamsAndPDs.iteritems():
113			if not self.perStreamPDList.has_key(stream): self.Results['reqStreamsAndPDs'].append(stream)
114			for PD in PDList:
115			if not PD in self.requiredStreamsAndPDs[stream]: self.Results['reqStreamsAndPDs'].append(stream+'::'+PD)
116
117			def checkExpress(self):
118			self.Results['checkExpress'] = []
119			if not self.perStreamPDList.has_key(self.ExpressStreamName):
120			self.Results['checkExpress'].append(self.ExpressStreamName)
121			return
122
123			if len(self.perStreamPDList[self.ExpressStreamName]) >1:
124			self.Results['checkExpress'].append("MULTIPLE_PDS")
125			if len(self.perStreamPDList[self.ExpressStreamName]) <1:
126			self.Results['checkExpress'].append("NO_PDS")
127
128			for PD in self.perStreamPDList[self.ExpressStreamName]:
129			if not self.expressPDs.has_key(PD):
130			self.Results['checkExpress'].append(self.ExpressStreamName+"::"+PD)
131			else:
132			self.expressType = self.expressPDs[PD]
133
134
135
136			def checkNameFormats(self):
137			self.Results['checkNameFormats']=[]
138			for PD in self.perPDPathList.iterkeys():
139			if not self.T0REGEXP['RXDATASET'].match(PD):
140			self.Results['checkNameFormats'].append(PD)
141			for path in self.perPathModuleList.iterkeys():
142			if not self.T0REGEXP['RXSAFEPATH'].match(path):
143			self.Results['checkNameFormats'].append(path)
144
145			def GetModules(self,cursor):
146			sqlquery ="""
147			SELECT I.NAME,E.NAME,D.NAME,I.ISENDPATH
148			FROM
149			CMS_HLT.PARAMETERS B,
150			CMS_HLT.SUPERIDPARAMETERASSOC C,
151			CMS_HLT.MODULETEMPLATES D,
152			CMS_HLT.MODULES E,
153			CMS_HLT.PATHMODULEASSOC F,
154			CMS_HLT.CONFIGURATIONPATHASSOC G,
155			CMS_HLT.CONFIGURATIONS H,
156			CMS_HLT.PATHS I
157			WHERE
158			B.PARAMID = C.PARAMID AND
159			C.SUPERID = F.MODULEID AND
160			E.TEMPLATEID = D.SUPERID AND
161			F.MODULEID = E.SUPERID AND
162			F.PATHID=G.PATHID AND
163			I.PATHID=G.PATHID AND
164			G.CONFIGID=H.CONFIGID AND
165			H.CONFIGDESCRIPTOR='%s'
166			""" % (self.menuName,)
167
168			cursor.execute(sqlquery)
169			for PathName,ModuleName,ModuleType,endPath in cursor.fetchall():
170			if not self.perPathModuleList.has_key(PathName): self.perPathModuleList[PathName] = []
171			self.perPathModuleList[PathName].append(ModuleName)
172			self.perModuleTypeList[ModuleName] = ModuleType
173			if endPath: self.endPathList.add(PathName)
174
175
176			def GetStreamsPathsPDs(self,cursor):
177			sqlquery= """
178			SELECT A.STREAMLABEL,E.NAME,F.DATASETLABEL
179			FROM
180			CMS_HLT.STREAMS A,
181			CMS_HLT.CONFIGURATIONS B,
182			CMS_HLT.CONFIGURATIONPATHASSOC C,
183			CMS_HLT.PATHSTREAMDATASETASSOC D,
184			CMS_HLT.PATHS E,
185			CMS_HLT.PRIMARYDATASETS F
186			WHERE
187			B.CONFIGDESCRIPTOR='%s' AND
188			C.CONFIGID=B.CONFIGID AND
189			D.PATHID=C.PATHID AND
190			A.STREAMID=D.STREAMID AND
191			E.PATHID = C.PATHID AND
192			F.DATASETID = D.DATASETID
193			""" % (self.menuName,)
194
195			cursor.execute(sqlquery)
196			for StreamName,PathName,PDName in cursor.fetchall():
197			if not self.perStreamPDList.has_key(StreamName): self.perStreamPDList[StreamName] = []
198			if not PDName in self.perStreamPDList[StreamName]: self.perStreamPDList[StreamName].append(PDName)
199			if not self.perPDPathList.has_key(PDName): self.perPDPathList[PDName] = []
200			self.perPDPathList[PDName].append(PathName)
201