CRAB/python/PhEDExDatasvcInfo.py

from Actor import *
import urllib
from xml.dom.minidom import parse
from crab_exceptions import *
from crab_logger import Logger
from WorkSpace import *
from urlparse import urlparse 
from LFNBaseName import *

class PhEDExDatasvcInfo:
    def __init__( self , cfg_params ):
 
        ## PhEDEx Data Service URL
        url="https://cmsweb.cern.ch/phedex/datasvc/xml/prod"
        self.datasvc_url = cfg_params.get("USER.datasvc_url",url)

        self.FacOps_savannah = 'https://savannah.cern.ch/support/?func=additem&group=cmscompinfrasup'


        self.srm_version = cfg_params.get("USER.srm_version",'srmv2')
        self.node = cfg_params.get('USER.storage_element',None)
        
        self.publish_data = cfg_params.get("USER.publish_data",0)
        self.usenamespace = cfg_params.get("USER.usenamespace",0)
        self.user_remote_dir = cfg_params.get("USER.user_remote_dir",'')
        if self.user_remote_dir:
            if ( self.user_remote_dir[-1] != '/' ) : self.user_remote_dir = self.user_remote_dir + '/'
            
        self.datasetpath = cfg_params.get("CMSSW.datasetpath")
        self.publish_data_name = cfg_params.get('USER.publish_data_name','')

        self.user_lfn = cfg_params.get("USER.lfn",'')
        if self.user_lfn:
            if ( self.user_lfn[-1] != '/' ) : self.user_lfn = self.user_lfn + '/'
            
        self.user_port = cfg_params.get("USER.storage_port",'8443')
        self.user_se_path = cfg_params.get("USER.storage_path",'')
        if self.user_se_path:
            if ( self.user_se_path[-1] != '/' ) : self.user_se_path = self.user_se_path + '/'
                                                    
       
        #check if using "private" Storage
        self.usePhedex = True 
        stage_out_faq='https://twiki.cern.ch/twiki/bin/view/CMS/SWGuideCrabFaq#How_to_store_output_with_CRAB_2'
        if not self.node :
            msg = 'Please specify the storage_element name in your crab.cfg section [USER].\n'
            msg +='      For further information please visit : %s'%stage_out_faq 
            raise CrabException(msg)
        if (self.node.find('T1_') + self.node.find('T2_')+self.node.find('T3_')) == -3: self.usePhedex = False 
        if not self.usePhedex and ( self.user_lfn == '' or self.user_se_path == '' ):
            msg = 'You are asking to stage out without using CMS Storage Name convention. In this case you \n' 
            msg += '      must specify both lfn and storage_path in the crab.cfg section [USER].\n '
            msg += '      For further information please visit : %s'%stage_out_faq
            raise CrabException(msg)
        self.sched = common.scheduler.name().upper() 
        self.protocol = self.srm_version
        if self.sched in ['CAF','LSF']:self.protocol = 'direct'

        self.forced_path = '/store/user/'
        if not self.usePhedex: 
            self.forced_path = self.user_lfn
        return
 
    def getEndpoint(self):   
        '''
        Return full SE endpoint and related infos
        '''
        self.lfn = self.getLFN()
 
        #extract the PFN for the given node,LFN,protocol
        endpoint = self.getStageoutPFN()
   
        #extract SE name an SE_PATH (needed for publication)
        SE, SE_PATH, User = self.splitEndpoint(endpoint)

        return endpoint, self.lfn , SE, SE_PATH, User         
       
    def splitEndpoint(self, endpoint):
        '''
        Return relevant infos from endpoint  
        '''
        SE = ''
        SE_PATH = ''
        USER = ''
        if self.usePhedex: 
            if self.protocol == 'direct':
                query=endpoint
                SE_PATH = endpoint
                SE = self.sched
            else: 
                url = 'http://'+endpoint.split('://')[1]
                # python > 2.4
                # SE = urlparse(url).hostname 
                scheme, host, path, params, query, fragment = urlparse(url)
                SE = host.split(':')[0]
                SE_PATH = endpoint.split(host)[1]
            USER = (query.split('user')[1]).split('/')[1]
        else:
            SE = self.node
            SE_PATH = self.user_se_path + self.user_lfn
            try:
                USER = (self.lfn.split('user')[1]).split('/')[1]
            except:
                pass

        return SE, SE_PATH, USER 
   

    def getLFN(self):
        """
        define the LFN composing the needed pieces
        """
        lfn = ''
        l_User = False
        if not self.usePhedex and (int(self.publish_data) == 0 and int(self.usenamespace) == 0) :
            ### add here check if user is trying to force a wrong LFN using a T2  TODO
            ## check if storage_name is a T2 (siteDB query)
            ## if yes :match self.user_lfn with LFNBaseName...
            ##     if NOT : raise (you are using a T2. It's not allowed stage out into self.user_path+self.user_lfn)   
            lfn = self.user_lfn
            return lfn
        if self.publish_data_name == '' and int(self.publish_data) == 1:
            msg = "Eeror. The [USER] section does not have 'publish_data_name'"
            raise CrabException(msg)
        if self.publish_data_name == '' and int(self.usenamespace) == 1:
           self.publish_data_name = "DefaultDataset"
        if int(self.publish_data) == 1 or int(self.usenamespace) == 1:
            if self.sched in ['CAF']: l_User=True 
            primaryDataset = self.computePrimaryDataset()
            lfn = LFNBase(self.forced_path, primaryDataset, self.publish_data_name, LocalUser=l_User)  + '/${PSETHASH}/'    
        else:
            if self.sched in ['CAF','LSF']: l_User=True 
            lfn = LFNBase(self.forced_path,self.user_remote_dir,LocalUser=l_User)
        return lfn
 
    def computePrimaryDataset(self):
        """
        compute the last part for the LFN in case of publication     
        """
        if (self.datasetpath.upper() != 'NONE'):
            primarydataset = self.datasetpath.split("/")[1]
        else:
            primarydataset = self.publish_data_name
        return primarydataset
    
    def lfn2pfn(self):
        """
        PhEDEx Data Service lfn2pfn call
 
        input:   LFN,node name,protocol
        returns: DOM object with the content of the PhEDEx Data Service call
        """  
        params = {'node' : self.node , 'lfn': self.lfn , 'protocol': self.protocol}
        params = urllib.urlencode(params)
        datasvc_lfn2pfn="%s/lfn2pfn"%self.datasvc_url
        urlresults = urllib.urlopen(datasvc_lfn2pfn, params)
        try:
            urlresults = parse(urlresults)
        except:
            urlresults = None

        return urlresults
 
    def parse_error(self,urlresults):
        """
        look for errors in the DOM object returned by PhEDEx Data Service call
        """
        errormsg = None 
        errors=urlresults.getElementsByTagName('error')
        for error in errors:
            errormsg=error.childNodes[0].data
            if len(error.childNodes)>1:
               errormsg+=error.childNodes[1].data
        return errormsg
 
    def parse_lfn2pfn(self,urlresults):
        """
        Parse the content of the result of lfn2pfn PhEDEx Data Service  call
 
        input:    DOM object with the content of the lfn2pfn call
        returns:  PFN  
        """
        result = urlresults.getElementsByTagName('phedex')
               
        if not result:
              return []
        result = result[0]
        pfn = None
        mapping = result.getElementsByTagName('mapping')
        for m in mapping:
            pfn=m.getAttribute("pfn")
            if pfn:
              return pfn
 
    def getStageoutPFN( self ):
        """
        input:   LFN,node name,protocol
        returns: PFN 
        """
        if self.usePhedex:
            fullurl="%s/lfn2pfn?node=%s&lfn=%s&protocol=%s"%(self.datasvc_url,self.node,self.lfn,self.protocol) 
            domlfn2pfn = self.lfn2pfn()
            if not domlfn2pfn :
                msg="Unable to get info from %s"%fullurl
                raise CrabException(msg)
  
            errormsg = self.parse_error(domlfn2pfn)
            if errormsg: 
                msg="Error extracting info from %s due to: %s"%(fullurl,errormsg)
                raise CrabException(msg)
  
            stageoutpfn = self.parse_lfn2pfn(domlfn2pfn)
            if not stageoutpfn:
                msg ='Unable to get stageout path from TFC at Site %s \n'%self.node
                msg+='      Please alert the CompInfraSup group through their savannah %s \n'%self.FacOps_savannah
                msg+='      reporting: \n'
                msg+='       Summary: Unable to get user stageout from TFC at Site %s \n'%self.node
                msg+='       OriginalSubmission: stageout path is not retrieved from %s \n'%fullurl
                raise CrabException(msg)
        else:
            if self.sched in ['CAF','LSF'] :
                stageoutpfn = self.user_se_path+self.lfn 
            else: 
                stageoutpfn = 'srm://'+self.node+':'+self.user_port+self.user_se_path+self.lfn 

        return stageoutpfn 


if __name__ == '__main__':
  """
  Sort of unit testing to check Phedex API for whatever site and/or lfn.
  Usage:
     python PhEDExDatasvcInfo.py --node T2_IT_Bari --lfn /store/maremma

  """
  import getopt,sys
  from crab_util import *
  import common
  klass_name = 'SchedulerGlite'
  klass = importName(klass_name, klass_name)
  common.scheduler = klass()

  lfn="/store/user/"
  node='T2_IT_Bari'
  valid = ['node=','lfn=']
  try:
       opts, args = getopt.getopt(sys.argv[1:], "", valid)
  except getopt.GetoptError, ex:
       print str(ex)
       sys.exit(1)
  for o, a in opts:
        if o == "--node":
            node = a
        if o == "--lfn":
            lfn = a
  
  mycfg_params = { 'USER.storage_element': node }
  dsvc = PhEDExDatasvcInfo(mycfg_params)
  dsvc.lfn = lfn
  print dsvc.getStageoutPFN()

Revision:	1.14
Committed:	Thu Oct 23 19:53:23 2008 UTC (16 years, 6 months ago) by spiga
Content type:	text/x-python
Branch:	MAIN
CVS Tags:	CRAB_2_4_2_pre2, CRAB_2_4_2_pre1, CRAB_2_4_1, CRAB_2_4_1_pre4
Changes since 1.13:	+5 -7 lines
Log Message:	minor fix
#	User	Rev	Content
1	spiga	1.3	from Actor import *
2	afanfani	1.1	import urllib
3			from xml.dom.minidom import parse
4			from crab_exceptions import *
5	spiga	1.3	from crab_logger import Logger
6			from WorkSpace import *
7			from urlparse import urlparse
8			from LFNBaseName import *
9	afanfani	1.1
10			class PhEDExDatasvcInfo:
11	spiga	1.3	def __init__( self , cfg_params ):
12
13			## PhEDEx Data Service URL
14			url="https://cmsweb.cern.ch/phedex/datasvc/xml/prod"
15			self.datasvc_url = cfg_params.get("USER.datasvc_url",url)
16
17	afanfani	1.6	self.FacOps_savannah = 'https://savannah.cern.ch/support/?func=additem&group=cmscompinfrasup'
18
19	spiga	1.3
20			self.srm_version = cfg_params.get("USER.srm_version",'srmv2')
21			self.node = cfg_params.get('USER.storage_element',None)
22
23			self.publish_data = cfg_params.get("USER.publish_data",0)
24			self.usenamespace = cfg_params.get("USER.usenamespace",0)
25	fanzago	1.10	self.user_remote_dir = cfg_params.get("USER.user_remote_dir",'')
26	fanzago	1.7	if self.user_remote_dir:
27			if ( self.user_remote_dir[-1] != '/' ) : self.user_remote_dir = self.user_remote_dir + '/'
28
29	spiga	1.3	self.datasetpath = cfg_params.get("CMSSW.datasetpath")
30			self.publish_data_name = cfg_params.get('USER.publish_data_name','')
31
32			self.user_lfn = cfg_params.get("USER.lfn",'')
33	fanzago	1.7	if self.user_lfn:
34			if ( self.user_lfn[-1] != '/' ) : self.user_lfn = self.user_lfn + '/'
35
36	spiga	1.8	self.user_port = cfg_params.get("USER.storage_port",'8443')
37	spiga	1.3	self.user_se_path = cfg_params.get("USER.storage_path",'')
38	fanzago	1.7	if self.user_se_path:
39			if ( self.user_se_path[-1] != '/' ) : self.user_se_path = self.user_se_path + '/'
40
41	spiga	1.3
42			#check if using "private" Storage
43			self.usePhedex = True
44	spiga	1.11	stage_out_faq='https://twiki.cern.ch/twiki/bin/view/CMS/SWGuideCrabFaq#How_to_store_output_with_CRAB_2'
45			if not self.node :
46			msg = 'Please specify the storage_element name in your crab.cfg section [USER].\n'
47			msg +=' For further information please visit : %s'%stage_out_faq
48			raise CrabException(msg)
49	spiga	1.4	if (self.node.find('T1_') + self.node.find('T2_')+self.node.find('T3_')) == -3: self.usePhedex = False
50	spiga	1.3	if not self.usePhedex and ( self.user_lfn == '' or self.user_se_path == '' ):
51			msg = 'You are asking to stage out without using CMS Storage Name convention. In this case you \n'
52			msg += ' must specify both lfn and storage_path in the crab.cfg section [USER].\n '
53	spiga	1.11	msg += ' For further information please visit : %s'%stage_out_faq
54	spiga	1.3	raise CrabException(msg)
55			self.sched = common.scheduler.name().upper()
56			self.protocol = self.srm_version
57			if self.sched in ['CAF','LSF']:self.protocol = 'direct'
58
59	spiga	1.14	self.forced_path = '/store/user/'
60			if not self.usePhedex:
61			self.forced_path = self.user_lfn
62	spiga	1.3	return
63
64			def getEndpoint(self):
65			'''
66			Return full SE endpoint and related infos
67			'''
68			self.lfn = self.getLFN()
69
70			#extract the PFN for the given node,LFN,protocol
71			endpoint = self.getStageoutPFN()
72
73			#extract SE name an SE_PATH (needed for publication)
74			SE, SE_PATH, User = self.splitEndpoint(endpoint)
75
76			return endpoint, self.lfn , SE, SE_PATH, User
77
78			def splitEndpoint(self, endpoint):
79			'''
80			Return relevant infos from endpoint
81			'''
82			SE = ''
83			SE_PATH = ''
84			USER = ''
85			if self.usePhedex:
86			if self.protocol == 'direct':
87			query=endpoint
88			SE_PATH = endpoint
89	fanzago	1.7	SE = self.sched
90	spiga	1.3	else:
91			url = 'http://'+endpoint.split('://')[1]
92			# python > 2.4
93			# SE = urlparse(url).hostname
94			scheme, host, path, params, query, fragment = urlparse(url)
95			SE = host.split(':')[0]
96			SE_PATH = endpoint.split(host)[1]
97			USER = (query.split('user')[1]).split('/')[1]
98			else:
99			SE = self.node
100			SE_PATH = self.user_se_path + self.user_lfn
101			try:
102			USER = (self.lfn.split('user')[1]).split('/')[1]
103			except:
104			pass
105
106			return SE, SE_PATH, USER
107	afanfani	1.2
108	spiga	1.3
109			def getLFN(self):
110			"""
111			define the LFN composing the needed pieces
112			"""
113			lfn = ''
114			l_User = False
115			if not self.usePhedex and (int(self.publish_data) == 0 and int(self.usenamespace) == 0) :
116			### add here check if user is trying to force a wrong LFN using a T2 TODO
117			## check if storage_name is a T2 (siteDB query)
118			## if yes :match self.user_lfn with LFNBaseName...
119			## if NOT : raise (you are using a T2. It's not allowed stage out into self.user_path+self.user_lfn)
120			lfn = self.user_lfn
121			return lfn
122			if self.publish_data_name == '' and int(self.publish_data) == 1:
123			msg = "Eeror. The [USER] section does not have 'publish_data_name'"
124			raise CrabException(msg)
125			if self.publish_data_name == '' and int(self.usenamespace) == 1:
126			self.publish_data_name = "DefaultDataset"
127			if int(self.publish_data) == 1 or int(self.usenamespace) == 1:
128			if self.sched in ['CAF']: l_User=True
129			primaryDataset = self.computePrimaryDataset()
130	spiga	1.14	lfn = LFNBase(self.forced_path, primaryDataset, self.publish_data_name, LocalUser=l_User) + '/${PSETHASH}/'
131	spiga	1.3	else:
132	spiga	1.5	if self.sched in ['CAF','LSF']: l_User=True
133	spiga	1.14	lfn = LFNBase(self.forced_path,self.user_remote_dir,LocalUser=l_User)
134	spiga	1.3	return lfn
135
136			def computePrimaryDataset(self):
137			"""
138			compute the last part for the LFN in case of publication
139			"""
140			if (self.datasetpath.upper() != 'NONE'):
141			primarydataset = self.datasetpath.split("/")[1]
142			else:
143			primarydataset = self.publish_data_name
144			return primarydataset
145
146			def lfn2pfn(self):
147			"""
148			PhEDEx Data Service lfn2pfn call
149
150			input: LFN,node name,protocol
151			returns: DOM object with the content of the PhEDEx Data Service call
152			"""
153			params = {'node' : self.node , 'lfn': self.lfn , 'protocol': self.protocol}
154			params = urllib.urlencode(params)
155			datasvc_lfn2pfn="%s/lfn2pfn"%self.datasvc_url
156			urlresults = urllib.urlopen(datasvc_lfn2pfn, params)
157			try:
158			urlresults = parse(urlresults)
159			except:
160			urlresults = None
161
162			return urlresults
163	afanfani	1.1
164	spiga	1.3	def parse_error(self,urlresults):
165			"""
166			look for errors in the DOM object returned by PhEDEx Data Service call
167			"""
168			errormsg = None
169			errors=urlresults.getElementsByTagName('error')
170			for error in errors:
171			errormsg=error.childNodes[0].data
172			if len(error.childNodes)>1:
173			errormsg+=error.childNodes[1].data
174			return errormsg
175
176			def parse_lfn2pfn(self,urlresults):
177			"""
178			Parse the content of the result of lfn2pfn PhEDEx Data Service call
179
180			input: DOM object with the content of the lfn2pfn call
181			returns: PFN
182			"""
183			result = urlresults.getElementsByTagName('phedex')
184
185			if not result:
186			return []
187			result = result[0]
188			pfn = None
189			mapping = result.getElementsByTagName('mapping')
190			for m in mapping:
191			pfn=m.getAttribute("pfn")
192			if pfn:
193			return pfn
194
195			def getStageoutPFN( self ):
196			"""
197			input: LFN,node name,protocol
198			returns: PFN
199			"""
200			if self.usePhedex:
201			fullurl="%s/lfn2pfn?node=%s&lfn=%s&protocol=%s"%(self.datasvc_url,self.node,self.lfn,self.protocol)
202			domlfn2pfn = self.lfn2pfn()
203			if not domlfn2pfn :
204			msg="Unable to get info from %s"%fullurl
205			raise CrabException(msg)
206
207			errormsg = self.parse_error(domlfn2pfn)
208			if errormsg:
209			msg="Error extracting info from %s due to: %s"%(fullurl,errormsg)
210			raise CrabException(msg)
211
212			stageoutpfn = self.parse_lfn2pfn(domlfn2pfn)
213			if not stageoutpfn:
214	afanfani	1.6	msg ='Unable to get stageout path from TFC at Site %s \n'%self.node
215			msg+=' Please alert the CompInfraSup group through their savannah %s \n'%self.FacOps_savannah
216			msg+=' reporting: \n'
217			msg+=' Summary: Unable to get user stageout from TFC at Site %s \n'%self.node
218			msg+=' OriginalSubmission: stageout path is not retrieved from %s \n'%fullurl
219	spiga	1.3	raise CrabException(msg)
220			else:
221	spiga	1.11	if self.sched in ['CAF','LSF'] :
222	spiga	1.12	stageoutpfn = self.user_se_path+self.lfn
223	spiga	1.11	else:
224			stageoutpfn = 'srm://'+self.node+':'+self.user_port+self.user_se_path+self.lfn
225	spiga	1.3
226			return stageoutpfn
227	afanfani	1.6
228
229
230			if __name__ == '__main__':
231			"""
232			Sort of unit testing to check Phedex API for whatever site and/or lfn.
233			Usage:
234			python PhEDExDatasvcInfo.py --node T2_IT_Bari --lfn /store/maremma
235
236			"""
237			import getopt,sys
238			from crab_util import *
239			import common
240			klass_name = 'SchedulerGlite'
241			klass = importName(klass_name, klass_name)
242			common.scheduler = klass()
243
244			lfn="/store/user/"
245			node='T2_IT_Bari'
246			valid = ['node=','lfn=']
247			try:
248			opts, args = getopt.getopt(sys.argv[1:], "", valid)
249			except getopt.GetoptError, ex:
250			print str(ex)
251			sys.exit(1)
252			for o, a in opts:
253			if o == "--node":
254			node = a
255			if o == "--lfn":
256			lfn = a
257
258			mycfg_params = { 'USER.storage_element': node }
259			dsvc = PhEDExDatasvcInfo(mycfg_params)
260			dsvc.lfn = lfn
261			print dsvc.getStageoutPFN()
262