ViewVC Help

View File | Revision Log | Show Annotations | Root Listing

root/cvsroot/COMP/CRAB/python/PhEDExDatasvcInfo.py

Comparing COMP/CRAB/python/PhEDExDatasvcInfo.py (file contents):
Revision 1.2 by afanfani, Fri Jul 18 10:00:54 2008 UTC vs.
Revision 1.41 by fanzago, Fri Aug 5 15:39:08 2011 UTC

#	Line 1 | Line 1
1	<	#!/usr/bin/env python
2	<
1	>	from Actor import *
2		import urllib
3		from xml.dom.minidom import parse
4		from crab_exceptions import *
5	+	from WorkSpace import *
6	+	from urlparse import urlparse
7	+	from LFNBaseName import *
8	+	from crab_util import getUserName
9
10		class PhEDExDatasvcInfo:
11	<	"""
12	<	provides information from PhEDEx Data Service
13	<	"""
14	<	## PhEDEx Data Service URL
15	<	#datasvc_url="https://cmsweb.cern.ch/phedex/test/datasvc/xml/prod"
16	<	datasvc_url="https://cmsweb.cern.ch/phedex/datasvc/xml/prod"
11	>	def __init__( self , cfg_params=None, config=None ):
12	>
13	>	## PhEDEx Data Service URL
14	>	self.datasvc_url="https://cmsweb.cern.ch/phedex/datasvc/xml/prod"
15	>
16	>	self.FacOps_savannah = 'https://savannah.cern.ch/support/?func=additem&group=cmscompinfrasup'
17	>	self.stage_out_faq='https://twiki.cern.ch/twiki/bin/view/CMS/SWGuideCrabHowTo#Stageout_and_publication'
18	>	self.dataPub_faq = 'https://twiki.cern.ch/twiki/bin/view/CMS/SWGuideCrabForPublication'
19	>
20	>	self.usePhedex = True
21	>	self.sched = common.scheduler.name().upper()
22	>
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	>	self.srm_version = cfg_params.get("USER.srm_version",'srmv2')
43	>	self.node = cfg_params.get('USER.storage_element',None)
44	>
45	>	self.publish_data = cfg_params.get("USER.publish_data",0)
46	>	self.usenamespace = cfg_params.get("USER.usenamespace",0)
47	>	self.user_remote_dir = cfg_params.get("USER.user_remote_dir",'')
48	>	if self.user_remote_dir:
49	>	if ( self.user_remote_dir[-1] != '/' ) : self.user_remote_dir = self.user_remote_dir + '/'
50	>
51	>	self.datasetpath = cfg_params.get("CMSSW.datasetpath")
52	>	self.publish_data_name = cfg_params.get('USER.publish_data_name','')
53	>
54	>	self.user_port = cfg_params.get("USER.storage_port",'8443')
55	>	self.user_se_path = cfg_params.get("USER.storage_path",'')
56	>	if self.user_se_path:
57	>	if ( self.user_se_path[-1] != '/' ) : self.user_se_path = self.user_se_path + '/'
58	>
59	>	#check if using "private" Storage
60	>	if not self.node :
61	>	msg = 'Please specify the storage_element name in your crab.cfg section [USER].\n'
62	>	msg +='\tFor further information please visit : %s'%self.stage_out_faq
63	>	raise CrabException(msg)
64	>	if (self.node.find('T1_') + self.node.find('T2_')+self.node.find('T3_')) == -3: self.usePhedex = False
65	>
66	>	if not self.usePhedex and ( self.user_remote_dir == '' or self.user_se_path == '' ):
67	>	####### FEDE FOR BUG 73010 ############
68	>	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	>	task = common._db.getTask()
72	>	add = '\n\n'
73	>	import shutil
74	>	try:
75	>	add += '      Task not correctly created: removing the working_dir ' +  common.work_space._top_dir + ' \n'
76	>	shutil.rmtree(common.work_space._top_dir)
77	>	except OSError:
78	>	add += '      Warning: problems removing the working_dir ' + common.work_space._top_dir + ' \n'
79	>	add += '      Please remove it by hand'
80	>	msg += add
81	>	raise CrabException(msg)
82	>	#########################################
83	>
84	>	self.forced_path = '/store/user/'
85	>	if self.sched in ['CAF','LSF','PBS']:
86	>	self.srm_version = 'direct'
87	>	self.SE = {'CAF':'caf.cern.ch', 'LSF':'', 'PBS':''}
88	>	if self.sched == 'CAF': self.forced_path = '/store/caf/user/'
89	>
90	>	if not self.usePhedex:
91	>	self.forced_path = self.user_remote_dir
92	>	return
93	>
94	>	def getEndpoint(self):
95	>	'''
96	>	Return full SE endpoint and related infos
97	>	'''
98	>	self.lfn = self.getLFN()
99	>
100	>	#extract the PFN for the given node,LFN,protocol
101	>	endpoint = self.getStageoutPFN()
102	>	if ( endpoint[-1] != '/' ) : endpoint = endpoint + '/'
103
104	<	def lfn2pfn(self,node,lfn,protocol):
105	<	"""
106	<	PhEDEx Data Service lfn2pfn call
107	<
108	<	input:   LFN,node name,protocol
109	<	returns: DOM object with the content of the PhEDEx Data Service call
110	<	"""
111	<	params = {'node' : node , 'lfn': lfn , 'protocol': protocol}
112	<	params = urllib.urlencode(params)
113	<	#print params
114	<	datasvc_lfn2pfn="%s/lfn2pfn"%self.datasvc_url
115	<	urlresults = urllib.urlopen(datasvc_lfn2pfn, params)
116	<	try:
117	<	urlresults = parse(urlresults)
118	<	except:
119	<	urlresults = None
120	<	return urlresults
121	<
122	<	def parse_error(self,urlresults):
123	<	"""
124	<	look for errors in the DOM object returned by PhEDEx Data Service call
125	<	"""
126	<	errormsg = None
127	<	errors=urlresults.getElementsByTagName('error')
128	<	for error in errors:
129	<	errormsg=error.childNodes[0].data
130	<	if len(error.childNodes)>1:
131	<	errormsg+=error.childNodes[1].data
132	<	return errormsg
133	<
134	<	def parse_lfn2pfn(self,urlresults):
135	<	"""
136	<	Parse the content of the result of lfn2pfn PhEDEx Data Service  call
137	<
138	<	input:    DOM object with the content of the lfn2pfn call
139	<	returns:  PFN
140	<	"""
141	<	result = urlresults.getElementsByTagName('phedex')
142	<	if not result:
143	<	return []
144	<	result = result[0]
145	<	pfn = None
146	<	mapping = result.getElementsByTagName('mapping')
147	<	for m in mapping:
148	<	pfn=m.getAttribute("pfn")
149	<	if pfn:
150	<	return pfn
151	<
152	<	def getStageoutPFN(self,node,lfn,protocol):
153	<	"""
154	<	input:   LFN,node name,protocol
155	<	returns: PFN
156	<	"""
157	<	fullurl="%s/lfn2pfn?node=%s&lfn=%s&protocol=%s"%(self.datasvc_url,node,lfn,protocol)
158	<	domlfn2pfn = self.lfn2pfn(node,lfn,protocol)
159	<	if not domlfn2pfn :
160	<	msg="Unable to get info from %s"%fullurl
161	<	raise CrabException(msg)
162	<
163	<	errormsg = self.parse_error(domlfn2pfn)
164	<	if errormsg:
165	<	msg="Error extracting info from %s due to: %s"%(fullurl,errormsg)
166	<	raise CrabException(msg)
167	<
168	<	stageoutpfn = self.parse_lfn2pfn(domlfn2pfn)
169	<	if not stageoutpfn:
170	<	msg="Unable to get stageout path (PFN) from %s"%fullurl
171	<	raise CrabException(msg)
172	<	return stageoutpfn
173	<
174	<
175	<	if __name__ == '__main__' :
176	<	"""
177	<	"""
178	<	from crab_logger import Logger
179	<	from WorkSpace import *
180	<	continue_dir="/home/fanfani/CRAB"
181	<	cfg_params={'USER.logdir' : continue_dir }
182	<	common.work_space = WorkSpace(continue_dir, cfg_params)
183	<	log = Logger()
95	<	common.logger = log
96	<
97	<	from LFNBaseName import *
98	<	# test values
99	<	lfn = LFNBase("datasetstring")
100	<	node='T2_IT_Bari'
101	<	protocol="srmv2"
102	<
103	<	#create an instance of the PhEDExDatasvcInfo object
104	<	dsvc = PhEDExDatasvcInfo()
105	<	#extract the PFN for the given node,LFN,protocol
106	<	print "Stageout to %s"%dsvc.getStageoutPFN(node,lfn,protocol)
104	>	#extract SE name an SE_PATH (needed for publication)
105	>	SE, SE_PATH, User = self.splitEndpoint(endpoint)
106	>
107	>	return endpoint, self.lfn , SE, SE_PATH, User
108	>
109	>	def splitEndpoint(self, endpoint):
110	>	'''
111	>	Return relevant infos from endpoint
112	>	'''
113	>	SE = ''
114	>	SE_PATH = ''
115	>	USER = getUserName()
116	>	if self.usePhedex:
117	>	if self.protocol == 'direct':
118	>	query=endpoint
119	>	SE_PATH = endpoint
120	>	SE = self.SE[self.sched]
121	>	else:
122	>	url = 'http://'+endpoint.split('://')[1]
123	>	scheme, host, path, params, query, fragment = urlparse(url)
124	>	SE = self.getAuthoritativeSE()
125	>	SE_PATH = endpoint.split(host)[1]
126	>	else:
127	>	SE = self.node
128	>	SE_PATH = self.user_se_path + self.user_remote_dir
129	>	if self.lfn.find('group') != -1:
130	>	try:
131	>	USER = (self.lfn.split('group')[1]).split('/')[1]
132	>	except:
133	>	pass
134	>	return SE, SE_PATH, USER
135	>
136	>	def getLFN(self):
137	>	"""
138	>	define the LFN composing the needed pieces
139	>	"""
140	>	lfn = ''
141	>	l_User = False
142	>	if not self.usePhedex and (int(self.publish_data) == 0 and int(self.usenamespace) == 0) :
143	>	### add here check if user is trying to force a wrong LFN using a T2  TODO
144	>	## check if storage_name is a T2 (siteDB query)
145	>	## if yes :match self.user_lfn with LFNBaseName...
146	>	##     if NOT : raise (you are using a T2. It's not allowed stage out into self.user_path+self.user_lfn)
147	>	lfn = self.user_remote_dir
148	>	return lfn
149	>	if self.publish_data_name == '' and int(self.publish_data) == 1:
150	>	msg = "Error. The [USER] section does not have 'publish_data_name'\n"
151	>	msg += '\tFor further information please visit : \n\t%s'%self.dataPub_faq
152	>	raise CrabException(msg)
153	>	if self.publish_data_name == '' and int(self.usenamespace) == 1:
154	>	self.publish_data_name = "DefaultDataset"
155	>	if int(self.publish_data) == 1:
156	>	if self.sched in ['CAF']: l_User=True
157	>	primaryDataset = self.computePrimaryDataset()
158	>	### added the case lfn = LFNBase(self.forced_path, primaryDataset, self.publish_data_name, publish=True)
159	>	### for the publication in order to be able to check the lfn length
160	>	lfn = LFNBase(self.forced_path, primaryDataset, self.publish_data_name, publish=True)  + '/${PSETHASH}/'
161	>	elif int(self.usenamespace) == 1:
162	>	if self.sched in ['CAF']: l_User=True
163	>	primaryDataset = self.computePrimaryDataset()
164	>	lfn = LFNBase(self.forced_path, primaryDataset, self.publish_data_name)  + '/${PSETHASH}/'
165	>	else:
166	>	if self.sched in ['CAF','LSF']: l_User=True
167	>	lfn = LFNBase(self.forced_path,self.user_remote_dir)
168	>	if ( lfn[-1] != '/' ) : lfn = lfn + '/'
169	>	return lfn
170	>
171	>	def computePrimaryDataset(self):
172	>	"""
173	>	compute the last part for the LFN in case of publication
174	>	"""
175	>	if (self.datasetpath.upper() != 'NONE'):
176	>	primarydataset = self.datasetpath.split("/")[1]
177	>	else:
178	>	primarydataset = self.publish_data_name
179	>	return primarydataset
180	>
181	>	def domPhedex(self,params,datasvc_baseUrl):
182	>	"""
183	>	PhEDEx Data Service lfn2pfn call
184
185	+	input:   params,datasvc_baseUrl
186	+	returns: DOM object with the content of the PhEDEx Data Service call
187	+	"""
188	+	params = urllib.urlencode(params)
189	+	try:
190	+	urlresults = urllib.urlopen(datasvc_baseUrl, params)
191	+	urlresults = parse(urlresults)
192	+	except IOError:
193	+	msg="Unable to access PhEDEx Data Service at %s"%datasvc_baseUrl
194	+	raise CrabException(msg)
195	+	except:
196	+	urlresults = None
197	+
198	+	return urlresults
199	+
200	+	def parse_error(self,urlresults):
201	+	"""
202	+	look for errors in the DOM object returned by PhEDEx Data Service call
203	+	"""
204	+	errormsg = None
205	+	errors=urlresults.getElementsByTagName('error')
206	+	for error in errors:
207	+	errormsg=error.childNodes[0].data
208	+	if len(error.childNodes)>1:
209	+	errormsg+=error.childNodes[1].data
210	+	return errormsg
211	+
212	+	def parse_lfn2pfn(self,urlresults):
213	+	"""
214	+	Parse the content of the result of lfn2pfn PhEDEx Data Service  call
215	+
216	+	input:    DOM object with the content of the lfn2pfn call
217	+	returns:  PFN
218	+	"""
219	+	result = urlresults.getElementsByTagName('phedex')
220	+
221	+	if not result:
222	+	return []
223	+	result = result[0]
224	+	pfn = None
225	+	mapping = result.getElementsByTagName('mapping')
226	+	for m in mapping:
227	+	pfn=m.getAttribute("pfn")
228	+	if pfn:
229	+	return pfn
230	+
231	+	def getStageoutPFN( self ):
232	+	"""
233	+	input:   LFN,node name,protocol
234	+	returns: PFN
235	+	"""
236	+	if self.usePhedex:
237	+	params = {'node' : self.node , 'lfn': self.lfn , 'protocol': self.protocol}
238	+	datasvc_lfn2pfn="%s/lfn2pfn"%self.datasvc_url
239	+	fullurl="%s/lfn2pfn?node=%s&lfn=%s&protocol=%s"%(self.datasvc_url,self.node,self.lfn,self.protocol)
240	+	domlfn2pfn = self.domPhedex(params,datasvc_lfn2pfn)
241	+	if not domlfn2pfn :
242	+	msg="Unable to get info from %s"%fullurl
243	+	raise CrabException(msg)
244	+
245	+	errormsg = self.parse_error(domlfn2pfn)
246	+	if errormsg:
247	+	msg="Error extracting info from %s due to: %s"%(fullurl,errormsg)
248	+	raise CrabException(msg)
249	+
250	+	stageoutpfn = self.parse_lfn2pfn(domlfn2pfn)
251	+	if not stageoutpfn:
252	+	msg ='Unable to get stageout path from TFC at Site %s \n'%self.node
253	+	msg+='      Please alert the CompInfraSup group through their savannah %s \n'%self.FacOps_savannah
254	+	msg+='      reporting: \n'
255	+	msg+='       Summary: Unable to get user stageout from TFC at Site %s \n'%self.node
256	+	msg+='       OriginalSubmission: stageout path is not retrieved from %s \n'%fullurl
257	+	raise CrabException(msg)
258	+	else:
259	+	if self.sched in ['CAF','LSF','PBS'] :
260	+	stageoutpfn = self.user_se_path+self.lfn
261	+	else:
262	+	stageoutpfn = 'srm://'+self.node+':'+self.user_port+self.user_se_path+self.lfn
263	+
264	+	if ( stageoutpfn[-1] != '/' ) : stageoutpfn = stageoutpfn + '/'
265	+	return stageoutpfn
266	+
267	+	def getAuthoritativeSE(self):
268	+	"""
269	+	input:   node name
270	+	returns: AuthoritativeSE
271	+	"""
272	+	params = {'node' : self.node }
273	+	datasvc_nodes="%s/nodes"%self.datasvc_url
274	+	fullurl="%s/nodes/?node=%s"%(self.datasvc_url,self.node)
275	+	domnodes = self.domPhedex(params,datasvc_nodes)
276	+
277	+	if not domnodes :
278	+	msg="Unable to get info from %s"%fullurl
279	+	raise CrabException(msg)
280	+
281	+	errormsg = self.parse_error(domnodes)
282	+	if errormsg:
283	+	msg="Error extracting info from %s due to: %s"%(fullurl,errormsg)
284	+	raise CrabException(msg)
285	+	result = domnodes.getElementsByTagName('phedex')
286	+	if not result:
287	+	return []
288	+	result = result[0]
289	+	se = None
290	+	node = result.getElementsByTagName('node')
291	+	for m in node:
292	+	se=m.getAttribute("se")
293	+	if se:
294	+	return se
295	+
296	+
297	+	if __name__ == '__main__':
298	+	"""
299	+	Sort of unit testing to check Phedex API for whatever site and/or lfn.
300	+	Usage:
301	+	python PhEDExDatasvcInfo.py --node T2_IT_Bari --lfn /store/maremma
302	+
303	+	"""
304	+	import getopt,sys
305	+	from crab_util import *
306	+	import common
307	+	klass_name = 'SchedulerGlite'
308	+	klass = importName(klass_name, klass_name)
309	+	common.scheduler = klass()
310	+
311	+	lfn="/store/user/"
312	+	node='T2_IT_Bari'
313	+	valid = ['node=','lfn=']
314	+	try:
315	+	opts, args = getopt.getopt(sys.argv[1:], "", valid)
316	+	except getopt.GetoptError, ex:
317	+	print str(ex)
318	+	sys.exit(1)
319	+	for o, a in opts:
320	+	if o == "--node":
321	+	node = a
322	+	if o == "--lfn":
323	+	lfn = a
324	+
325	+	mycfg_params = { 'USER.storage_element': node }
326	+	dsvc = PhEDExDatasvcInfo(mycfg_params)
327	+	dsvc.lfn = lfn
328	+	print dsvc.getStageoutPFN()
329	+

Diff Legend

–	Removed lines
+	Added lines
<	Changed lines
>	Changed lines