CRAB/python/PhEDExDatasvcInfo.py

from Actor import *
import urllib
from xml.dom.minidom import parse
from crab_exceptions import *
from WorkSpace import *
from urlparse import urlparse 
from LFNBaseName import *
from crab_util import getUserName

class PhEDExDatasvcInfo:
    def __init__( self , cfg_params=None, config=None ):
 
        ## PhEDEx Data Service URL
        self.datasvc_url="https://cmsweb.cern.ch/phedex/datasvc/xml/prod"

        self.FacOps_savannah = 'https://savannah.cern.ch/support/?func=additem&group=cmscompinfrasup'
        self.stage_out_faq='https://twiki.cern.ch/twiki/bin/view/CMS/SWGuideCrabHowTo#Stageout_and_publication' 
        self.dataPub_faq = 'https://twiki.cern.ch/twiki/bin/view/CMS/SWGuideCrabForPublication'

        self.usePhedex = True 
        self.sched = common.scheduler.name().upper() 

        if config!=None:
            self.checkConfig(config)  
        else:
            self.checkCfgConfig(cfg_params)  

        self.protocol = self.srm_version
 

    def checkConfig(self,config):
        """
        """
        self.srm_version = config.get("srm_version",'srmv2')
        self.node = config.get('storage_element',None)
        self.lfn='/store/'

    def checkCfgConfig(self,cfg_params):
        """
        """
        self.datasvc_url = cfg_params.get("USER.datasvc_url",self.datasvc_url)
        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_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
        if not self.node :
            msg = 'Please specify the storage_element name in your crab.cfg section [USER].\n'
            msg +='\tFor further information please visit : %s'%self.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_remote_dir == '' or self.user_se_path == '' ):
            ####### FEDE FOR BUG 73010 ############
            msg = 'Error: you are asking to stage out without using CMS Storage Name convention. In this case you \n' 
            msg += '      must specify both user_remote_dir and storage_path in the crab.cfg section [USER].\n'
            msg += '      For further information please visit : \n\t%s'%self.stage_out_faq
            task = common._db.getTask()
            add = '\n\n'
            import shutil
            try:
                add += '      Task not correctly created: removing the working_dir ' +  common.work_space._top_dir + ' \n'
                shutil.rmtree(common.work_space._top_dir)
            except OSError:
                add += '      Warning: problems removing the working_dir ' + common.work_space._top_dir + ' \n'
                add += '      Please remove it by hand'
            msg += add
            raise CrabException(msg)

        self.forced_path = '/store/user/'
        if self.sched in ['LSF','PBS']:
            self.srm_version = 'direct'
            self.SE = {'LSF':'', 'PBS':''}
            
        if self.sched == 'CAF': 
            #### FEDE TEST FOR XROOTD
            ######### first solution ################
            #eos = cfg_params.get("USER.caf_eos_area", 0)
            #if eos == 0:
            #    self.forced_path = '/store/caf/user/'
            #else:    
            #    self.forced_path = '/store/eos/user'
            #########################################
            ######### second solution ###############
            self.forced_path = cfg_params.get("USER.caf_lfn", '/store/caf/user')
            #########################################
            print "--->>> FORCING THE FIRST PART OF LFN WITH ", self.forced_path
            self.SE = {'CAF':'caf.cern.ch'}
            self.srm_version = 'stageout'
            print "--->>> query with 'stageout' "
            #########################################

        if not self.usePhedex: 
            self.forced_path = self.user_remote_dir
        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()
        if ( endpoint[-1] != '/' ) : endpoint = endpoint + '/'
   
        #extract SE name an SE_PATH (needed for publication)
        SE, SE_PATH, User = self.splitEndpoint(endpoint)

        #### FEDE FOR XROOTD #####
        #print "in getEndpoint di PhEDExDatasvcInfo.py: "
        #print "    SE = ", SE
        #print "    SE_PATH = ", SE_PATH
        #print "    User = ", User
        #print "    endpoint = ", endpoint
        ##############################

        return endpoint, self.lfn , SE, SE_PATH, User         
       
    def splitEndpoint(self, endpoint):
        '''
        Return relevant infos from endpoint  
        '''
        SE = ''
        SE_PATH = ''
        USER = getUserName()
        if self.usePhedex: 
            ### FEDE PER TEST WITH XROOTD
            if (self.protocol == 'direct' or self.protocol == 'stageout'):
                SE = self.SE[self.sched]
                SE_PATH = endpoint 
                #############################   
                #print "    SE_PATH = ", SE_PATH
            else: 
                url = 'http://'+endpoint.split('://')[1]
                scheme, host, path, params, query, fragment = urlparse(url)
                SE = self.getAuthoritativeSE()
                SE_PATH = endpoint.split(host)[1]
        else:
            SE = self.node
            SE_PATH = self.user_se_path + self.user_remote_dir 
            if self.lfn.find('group') != -1:
                try:
                    USER = (self.lfn.split('group')[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_remote_dir
            return lfn
        if self.publish_data_name == '' and int(self.publish_data) == 1:
            msg = "Error. The [USER] section does not have 'publish_data_name'\n"
            msg += '\tFor further information please visit : \n\t%s'%self.dataPub_faq
            raise CrabException(msg)
        if self.publish_data_name == '' and int(self.usenamespace) == 1:
           self.publish_data_name = "DefaultDataset"
        if int(self.publish_data) == 1:
            if self.sched in ['CAF']: l_User=True 
            primaryDataset = self.computePrimaryDataset()
            ### added the case lfn = LFNBase(self.forced_path, primaryDataset, self.publish_data_name, publish=True)
            ### for the publication in order to be able to check the lfn length  
            lfn = LFNBase(self.forced_path, primaryDataset, self.publish_data_name, publish=True)  + '/${PSETHASH}/'    
        elif int(self.usenamespace) == 1:
            if self.sched in ['CAF']: l_User=True 
            primaryDataset = self.computePrimaryDataset()
            lfn = LFNBase(self.forced_path, primaryDataset, self.publish_data_name)  + '/${PSETHASH}/'    
        else:
            if self.sched in ['CAF','LSF']: l_User=True 
            lfn = LFNBase(self.forced_path,self.user_remote_dir)
        if ( lfn[-1] != '/' ) : lfn = lfn + '/'
        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 domPhedex(self,params,datasvc_baseUrl):
        """
        PhEDEx Data Service lfn2pfn call
 
        input:   params,datasvc_baseUrl
        returns: DOM object with the content of the PhEDEx Data Service call
        """  
        params = urllib.urlencode(params)
        try:
            urlresults = urllib.urlopen(datasvc_baseUrl, params)
            urlresults = parse(urlresults)
        except IOError:
            msg="Unable to access PhEDEx Data Service at %s"%datasvc_baseUrl
            raise CrabException(msg)
        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:
            params = {'node' : self.node , 'lfn': self.lfn , 'protocol': self.protocol}
            datasvc_lfn2pfn="%s/lfn2pfn"%self.datasvc_url
            fullurl="%s/lfn2pfn?node=%s&lfn=%s&protocol=%s"%(self.datasvc_url,self.node,self.lfn,self.protocol) 
            print "--->>> fullurl = ", fullurl
            domlfn2pfn = self.domPhedex(params,datasvc_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','PBS'] :
                stageoutpfn = self.user_se_path+self.lfn 
            else: 
                stageoutpfn = 'srm://'+self.node+':'+self.user_port+self.user_se_path+self.lfn 

        if ( stageoutpfn[-1] != '/' ) : stageoutpfn = stageoutpfn + '/'
        return stageoutpfn 

    def getAuthoritativeSE(self):
        """
        input:   node name
        returns: AuthoritativeSE 
        """
        params = {'node' : self.node }
        datasvc_nodes="%s/nodes"%self.datasvc_url
        fullurl="%s/nodes/?node=%s"%(self.datasvc_url,self.node) 
        domnodes = self.domPhedex(params,datasvc_nodes)

        if not domnodes :
            msg="Unable to get info from %s"%fullurl
            raise CrabException(msg)

        errormsg = self.parse_error(domnodes)
        if errormsg: 
            msg="Error extracting info from %s due to: %s"%(fullurl,errormsg)
            raise CrabException(msg)
        result = domnodes.getElementsByTagName('phedex')
        if not result:
              return []
        result = result[0]
        se = None
        node = result.getElementsByTagName('node')
        for m in node:
            se=m.getAttribute("se")
            if se:
                return se


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:	1.42
Committed:	Mon Aug 29 09:48:03 2011 UTC (13 years, 8 months ago) by fanzago
Content type:	text/x-python
Branch:	MAIN
CVS Tags:	CRAB_2_7_9_pre2
Changes since 1.41:	+34 -7 lines
Log Message:	changes about xrootd plugin, savannah #83930
#	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 WorkSpace import *
6			from urlparse import urlparse
7			from LFNBaseName import *
8	fanzago	1.36	from crab_util import getUserName
9	afanfani	1.1
10			class PhEDExDatasvcInfo:
11	fanzago	1.28	def __init__( self , cfg_params=None, config=None ):
12	spiga	1.3
13			## PhEDEx Data Service URL
14	belforte	1.39	self.datasvc_url="https://cmsweb.cern.ch/phedex/datasvc/xml/prod"
15	spiga	1.3
16	afanfani	1.6	self.FacOps_savannah = 'https://savannah.cern.ch/support/?func=additem&group=cmscompinfrasup'
17	spiga	1.34	self.stage_out_faq='https://twiki.cern.ch/twiki/bin/view/CMS/SWGuideCrabHowTo#Stageout_and_publication'
18	spiga	1.17	self.dataPub_faq = 'https://twiki.cern.ch/twiki/bin/view/CMS/SWGuideCrabForPublication'
19	spiga	1.3
20	fanzago	1.28	self.usePhedex = True
21	spiga	1.29	self.sched = common.scheduler.name().upper()
22	fanzago	1.28
23			if config!=None:
24			self.checkConfig(config)
25			else:
26			self.checkCfgConfig(cfg_params)
27
28			self.protocol = self.srm_version
29
30
31			def checkConfig(self,config):
32			"""
33			"""
34			self.srm_version = config.get("srm_version",'srmv2')
35			self.node = config.get('storage_element',None)
36			self.lfn='/store/'
37
38			def checkCfgConfig(self,cfg_params):
39			"""
40			"""
41			self.datasvc_url = cfg_params.get("USER.datasvc_url",self.datasvc_url)
42	spiga	1.3	self.srm_version = cfg_params.get("USER.srm_version",'srmv2')
43			self.node = cfg_params.get('USER.storage_element',None)
44	spiga	1.17
45	spiga	1.3	self.publish_data = cfg_params.get("USER.publish_data",0)
46			self.usenamespace = cfg_params.get("USER.usenamespace",0)
47	fanzago	1.10	self.user_remote_dir = cfg_params.get("USER.user_remote_dir",'')
48	fanzago	1.7	if self.user_remote_dir:
49			if ( self.user_remote_dir[-1] != '/' ) : self.user_remote_dir = self.user_remote_dir + '/'
50	spiga	1.17
51	spiga	1.3	self.datasetpath = cfg_params.get("CMSSW.datasetpath")
52			self.publish_data_name = cfg_params.get('USER.publish_data_name','')
53
54	spiga	1.8	self.user_port = cfg_params.get("USER.storage_port",'8443')
55	spiga	1.3	self.user_se_path = cfg_params.get("USER.storage_path",'')
56	fanzago	1.7	if self.user_se_path:
57			if ( self.user_se_path[-1] != '/' ) : self.user_se_path = self.user_se_path + '/'
58
59	spiga	1.3	#check if using "private" Storage
60	spiga	1.11	if not self.node :
61			msg = 'Please specify the storage_element name in your crab.cfg section [USER].\n'
62	spiga	1.34	msg +='\tFor further information please visit : %s'%self.stage_out_faq
63	spiga	1.11	raise CrabException(msg)
64	spiga	1.4	if (self.node.find('T1_') + self.node.find('T2_')+self.node.find('T3_')) == -3: self.usePhedex = False
65	spiga	1.15
66			if not self.usePhedex and ( self.user_remote_dir == '' or self.user_se_path == '' ):
67	fanzago	1.40	####### FEDE FOR BUG 73010 ############
68	fanzago	1.41	msg = 'Error: you are asking to stage out without using CMS Storage Name convention. In this case you \n'
69			msg += ' must specify both user_remote_dir and storage_path in the crab.cfg section [USER].\n'
70			msg += ' For further information please visit : \n\t%s'%self.stage_out_faq
71	fanzago	1.40	task = common._db.getTask()
72	fanzago	1.41	add = '\n\n'
73	fanzago	1.40	import shutil
74			try:
75	fanzago	1.41	add += ' Task not correctly created: removing the working_dir ' + common.work_space._top_dir + ' \n'
76	fanzago	1.40	shutil.rmtree(common.work_space._top_dir)
77			except OSError:
78	fanzago	1.41	add += ' Warning: problems removing the working_dir ' + common.work_space._top_dir + ' \n'
79			add += ' Please remove it by hand'
80	fanzago	1.40	msg += add
81	spiga	1.3	raise CrabException(msg)
82	spiga	1.22
83			self.forced_path = '/store/user/'
84	fanzago	1.42	if self.sched in ['LSF','PBS']:
85	spiga	1.31	self.srm_version = 'direct'
86	fanzago	1.42	self.SE = {'LSF':'', 'PBS':''}
87	spiga	1.22
88	fanzago	1.42	if self.sched == 'CAF':
89			#### FEDE TEST FOR XROOTD
90			######### first solution ################
91			#eos = cfg_params.get("USER.caf_eos_area", 0)
92			#if eos == 0:
93			# self.forced_path = '/store/caf/user/'
94			#else:
95			# self.forced_path = '/store/eos/user'
96			#########################################
97			######### second solution ###############
98			self.forced_path = cfg_params.get("USER.caf_lfn", '/store/caf/user')
99			#########################################
100			print "--->>> FORCING THE FIRST PART OF LFN WITH ", self.forced_path
101			self.SE = {'CAF':'caf.cern.ch'}
102			self.srm_version = 'stageout'
103			print "--->>> query with 'stageout' "
104			#########################################
105
106	spiga	1.14	if not self.usePhedex:
107	spiga	1.15	self.forced_path = self.user_remote_dir
108	spiga	1.3	return
109
110			def getEndpoint(self):
111			'''
112			Return full SE endpoint and related infos
113			'''
114			self.lfn = self.getLFN()
115
116			#extract the PFN for the given node,LFN,protocol
117			endpoint = self.getStageoutPFN()
118	fanzago	1.27	if ( endpoint[-1] != '/' ) : endpoint = endpoint + '/'
119	spiga	1.3
120			#extract SE name an SE_PATH (needed for publication)
121			SE, SE_PATH, User = self.splitEndpoint(endpoint)
122
123	fanzago	1.42	#### FEDE FOR XROOTD #####
124			#print "in getEndpoint di PhEDExDatasvcInfo.py: "
125			#print " SE = ", SE
126			#print " SE_PATH = ", SE_PATH
127			#print " User = ", User
128			#print " endpoint = ", endpoint
129			##############################
130
131	spiga	1.3	return endpoint, self.lfn , SE, SE_PATH, User
132
133			def splitEndpoint(self, endpoint):
134			'''
135			Return relevant infos from endpoint
136			'''
137			SE = ''
138			SE_PATH = ''
139	spiga	1.37	USER = getUserName()
140	spiga	1.3	if self.usePhedex:
141	fanzago	1.42	### FEDE PER TEST WITH XROOTD
142			if (self.protocol == 'direct' or self.protocol == 'stageout'):
143	fanzago	1.18	SE = self.SE[self.sched]
144	fanzago	1.42	SE_PATH = endpoint
145			#############################
146			#print " SE_PATH = ", SE_PATH
147	spiga	1.3	else:
148			url = 'http://'+endpoint.split('://')[1]
149			scheme, host, path, params, query, fragment = urlparse(url)
150	spiga	1.33	SE = self.getAuthoritativeSE()
151	spiga	1.3	SE_PATH = endpoint.split(host)[1]
152			else:
153	spiga	1.20	SE = self.node
154			SE_PATH = self.user_se_path + self.user_remote_dir
155	spiga	1.37	if self.lfn.find('group') != -1:
156			try:
157			USER = (self.lfn.split('group')[1]).split('/')[1]
158			except:
159			pass
160	spiga	1.3	return SE, SE_PATH, USER
161
162			def getLFN(self):
163			"""
164			define the LFN composing the needed pieces
165			"""
166			lfn = ''
167			l_User = False
168			if not self.usePhedex and (int(self.publish_data) == 0 and int(self.usenamespace) == 0) :
169			### add here check if user is trying to force a wrong LFN using a T2 TODO
170			## check if storage_name is a T2 (siteDB query)
171			## if yes :match self.user_lfn with LFNBaseName...
172			## if NOT : raise (you are using a T2. It's not allowed stage out into self.user_path+self.user_lfn)
173	spiga	1.15	lfn = self.user_remote_dir
174	spiga	1.3	return lfn
175			if self.publish_data_name == '' and int(self.publish_data) == 1:
176	spiga	1.17	msg = "Error. The [USER] section does not have 'publish_data_name'\n"
177			msg += '\tFor further information please visit : \n\t%s'%self.dataPub_faq
178	spiga	1.3	raise CrabException(msg)
179			if self.publish_data_name == '' and int(self.usenamespace) == 1:
180			self.publish_data_name = "DefaultDataset"
181	fanzago	1.26	if int(self.publish_data) == 1:
182			if self.sched in ['CAF']: l_User=True
183			primaryDataset = self.computePrimaryDataset()
184			### added the case lfn = LFNBase(self.forced_path, primaryDataset, self.publish_data_name, publish=True)
185			### for the publication in order to be able to check the lfn length
186			lfn = LFNBase(self.forced_path, primaryDataset, self.publish_data_name, publish=True) + '/${PSETHASH}/'
187			elif int(self.usenamespace) == 1:
188	spiga	1.3	if self.sched in ['CAF']: l_User=True
189			primaryDataset = self.computePrimaryDataset()
190	spiga	1.21	lfn = LFNBase(self.forced_path, primaryDataset, self.publish_data_name) + '/${PSETHASH}/'
191	spiga	1.3	else:
192	spiga	1.5	if self.sched in ['CAF','LSF']: l_User=True
193	spiga	1.21	lfn = LFNBase(self.forced_path,self.user_remote_dir)
194	fanzago	1.35	if ( lfn[-1] != '/' ) : lfn = lfn + '/'
195	spiga	1.3	return lfn
196
197			def computePrimaryDataset(self):
198			"""
199			compute the last part for the LFN in case of publication
200			"""
201			if (self.datasetpath.upper() != 'NONE'):
202			primarydataset = self.datasetpath.split("/")[1]
203			else:
204			primarydataset = self.publish_data_name
205			return primarydataset
206
207	spiga	1.33	def domPhedex(self,params,datasvc_baseUrl):
208	spiga	1.3	"""
209			PhEDEx Data Service lfn2pfn call
210
211	spiga	1.33	input: params,datasvc_baseUrl
212	spiga	1.3	returns: DOM object with the content of the PhEDEx Data Service call
213			"""
214			params = urllib.urlencode(params)
215			try:
216	spiga	1.33	urlresults = urllib.urlopen(datasvc_baseUrl, params)
217	spiga	1.3	urlresults = parse(urlresults)
218	slacapra	1.24	except IOError:
219	spiga	1.33	msg="Unable to access PhEDEx Data Service at %s"%datasvc_baseUrl
220	slacapra	1.24	raise CrabException(msg)
221	spiga	1.3	except:
222			urlresults = None
223
224			return urlresults
225	afanfani	1.1
226	spiga	1.3	def parse_error(self,urlresults):
227			"""
228			look for errors in the DOM object returned by PhEDEx Data Service call
229			"""
230			errormsg = None
231			errors=urlresults.getElementsByTagName('error')
232			for error in errors:
233			errormsg=error.childNodes[0].data
234			if len(error.childNodes)>1:
235			errormsg+=error.childNodes[1].data
236			return errormsg
237
238			def parse_lfn2pfn(self,urlresults):
239			"""
240			Parse the content of the result of lfn2pfn PhEDEx Data Service call
241
242			input: DOM object with the content of the lfn2pfn call
243			returns: PFN
244			"""
245			result = urlresults.getElementsByTagName('phedex')
246
247			if not result:
248			return []
249			result = result[0]
250			pfn = None
251			mapping = result.getElementsByTagName('mapping')
252			for m in mapping:
253			pfn=m.getAttribute("pfn")
254			if pfn:
255			return pfn
256
257			def getStageoutPFN( self ):
258			"""
259			input: LFN,node name,protocol
260			returns: PFN
261			"""
262			if self.usePhedex:
263	spiga	1.33	params = {'node' : self.node , 'lfn': self.lfn , 'protocol': self.protocol}
264			datasvc_lfn2pfn="%s/lfn2pfn"%self.datasvc_url
265	spiga	1.3	fullurl="%s/lfn2pfn?node=%s&lfn=%s&protocol=%s"%(self.datasvc_url,self.node,self.lfn,self.protocol)
266	fanzago	1.42	print "--->>> fullurl = ", fullurl
267	spiga	1.33	domlfn2pfn = self.domPhedex(params,datasvc_lfn2pfn)
268	spiga	1.3	if not domlfn2pfn :
269			msg="Unable to get info from %s"%fullurl
270			raise CrabException(msg)
271
272			errormsg = self.parse_error(domlfn2pfn)
273			if errormsg:
274			msg="Error extracting info from %s due to: %s"%(fullurl,errormsg)
275			raise CrabException(msg)
276
277			stageoutpfn = self.parse_lfn2pfn(domlfn2pfn)
278			if not stageoutpfn:
279	afanfani	1.6	msg ='Unable to get stageout path from TFC at Site %s \n'%self.node
280			msg+=' Please alert the CompInfraSup group through their savannah %s \n'%self.FacOps_savannah
281			msg+=' reporting: \n'
282			msg+=' Summary: Unable to get user stageout from TFC at Site %s \n'%self.node
283			msg+=' OriginalSubmission: stageout path is not retrieved from %s \n'%fullurl
284	spiga	1.3	raise CrabException(msg)
285			else:
286	mcinquil	1.32	if self.sched in ['CAF','LSF','PBS'] :
287	spiga	1.12	stageoutpfn = self.user_se_path+self.lfn
288	spiga	1.11	else:
289			stageoutpfn = 'srm://'+self.node+':'+self.user_port+self.user_se_path+self.lfn
290	spiga	1.3
291	fanzago	1.35	if ( stageoutpfn[-1] != '/' ) : stageoutpfn = stageoutpfn + '/'
292	spiga	1.3	return stageoutpfn
293	afanfani	1.6
294	spiga	1.33	def getAuthoritativeSE(self):
295			"""
296			input: node name
297			returns: AuthoritativeSE
298			"""
299			params = {'node' : self.node }
300			datasvc_nodes="%s/nodes"%self.datasvc_url
301			fullurl="%s/nodes/?node=%s"%(self.datasvc_url,self.node)
302			domnodes = self.domPhedex(params,datasvc_nodes)
303
304			if not domnodes :
305			msg="Unable to get info from %s"%fullurl
306			raise CrabException(msg)
307
308			errormsg = self.parse_error(domnodes)
309			if errormsg:
310			msg="Error extracting info from %s due to: %s"%(fullurl,errormsg)
311			raise CrabException(msg)
312			result = domnodes.getElementsByTagName('phedex')
313			if not result:
314			return []
315			result = result[0]
316			se = None
317			node = result.getElementsByTagName('node')
318			for m in node:
319			se=m.getAttribute("se")
320			if se:
321			return se
322	afanfani	1.6
323
324			if __name__ == '__main__':
325			"""
326			Sort of unit testing to check Phedex API for whatever site and/or lfn.
327			Usage:
328			python PhEDExDatasvcInfo.py --node T2_IT_Bari --lfn /store/maremma
329
330			"""
331			import getopt,sys
332			from crab_util import *
333			import common
334			klass_name = 'SchedulerGlite'
335			klass = importName(klass_name, klass_name)
336			common.scheduler = klass()
337
338			lfn="/store/user/"
339			node='T2_IT_Bari'
340			valid = ['node=','lfn=']
341			try:
342			opts, args = getopt.getopt(sys.argv[1:], "", valid)
343			except getopt.GetoptError, ex:
344			print str(ex)
345			sys.exit(1)
346			for o, a in opts:
347			if o == "--node":
348			node = a
349			if o == "--lfn":
350			lfn = a
351
352			mycfg_params = { 'USER.storage_element': node }
353			dsvc = PhEDExDatasvcInfo(mycfg_params)
354			dsvc.lfn = lfn
355			print dsvc.getStageoutPFN()
356