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 1.1 by afanfani, Wed Jun 25 12:03:43 2008 UTC vs.
Revision 1.27 by fanzago, Wed Sep 16 15:49:01 2009 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
9		class PhEDExDatasvcInfo:
10	<	"""
11	<	provides information from PhEDEx Data Service
12	<	"""
13	<	# PhEDEx Data Service URL
14	<	datasvc_url="https://cmsweb.cern.ch/phedex/test/datasvc/xml/prod"
10	>	def __init__( self , cfg_params ):
11	>
12	>	## PhEDEx Data Service URL
13	>	url="https://cmsweb.cern.ch/phedex/datasvc/xml/prod"
14	>	self.datasvc_url = cfg_params.get("USER.datasvc_url",url)
15	>
16	>	self.FacOps_savannah = 'https://savannah.cern.ch/support/?func=additem&group=cmscompinfrasup'
17	>	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.srm_version = cfg_params.get("USER.srm_version",'srmv2')
21	>	self.node = cfg_params.get('USER.storage_element',None)
22	>
23	>
24	>
25	>	self.user_lfn = cfg_params.get("USER.lfn",'')
26	>	self.publish_data = cfg_params.get("USER.publish_data",0)
27	>	self.usenamespace = cfg_params.get("USER.usenamespace",0)
28	>	self.user_remote_dir = cfg_params.get("USER.user_remote_dir",'')
29	>	if self.user_remote_dir:
30	>	if ( self.user_remote_dir[-1] != '/' ) : self.user_remote_dir = self.user_remote_dir + '/'
31	>	if self.user_lfn:
32	>	msg  = 'Warning: lfn has been deprecated, CRAB will ignore it.\n'
33	>	msg += '\t Please use only user_remote_dir removing lfn from your crab.cfg\n'
34	>	msg += '\t For further information please visit : \n\t%s'%stage_out_faq
35	>	common.logger.info(msg)
36	>
37	>	self.datasetpath = cfg_params.get("CMSSW.datasetpath")
38	>	self.publish_data_name = cfg_params.get('USER.publish_data_name','')
39	>
40	>	self.user_port = cfg_params.get("USER.storage_port",'8443')
41	>	self.user_se_path = cfg_params.get("USER.storage_path",'')
42	>	if self.user_se_path:
43	>	if ( self.user_se_path[-1] != '/' ) : self.user_se_path = self.user_se_path + '/'
44	>
45	>	#check if using "private" Storage
46	>	self.usePhedex = True
47	>	if not self.node :
48	>	msg = 'Please specify the storage_element name in your crab.cfg section [USER].\n'
49	>	msg +='\tFor further information please visit : %s'%stage_out_faq
50	>	raise CrabException(msg)
51	>	if (self.node.find('T1_') + self.node.find('T2_')+self.node.find('T3_')) == -3: self.usePhedex = False
52	>
53	>	if not self.usePhedex and ( self.user_remote_dir == '' or self.user_se_path == '' ):
54	>	msg = 'You are asking to stage out without using CMS Storage Name convention. In this case you \n'
55	>	msg += '\t must specify both user_remote_dir and storage_path in the crab.cfg section [USER].\n '
56	>	msg += '\t For further information please visit : \n\t%s'%stage_out_faq
57	>	raise CrabException(msg)
58	>	self.sched = common.scheduler.name().upper()
59	>	self.protocol = self.srm_version
60	>
61	>	self.forced_path = '/store/user/'
62	>	if self.sched in ['CAF','LSF']:
63	>	self.protocol = 'direct'
64	>	self.SE = {'CAF':'caf.cern.ch', 'LSF':''}
65	>	if self.sched == 'CAF': self.forced_path = '/store/caf/user/'
66	>
67	>	if not self.usePhedex:
68	>	self.forced_path = self.user_remote_dir
69	>	return
70	>
71	>	def getEndpoint(self):
72	>	'''
73	>	Return full SE endpoint and related infos
74	>	'''
75	>	self.lfn = self.getLFN()
76	>
77	>	#extract the PFN for the given node,LFN,protocol
78	>	endpoint = self.getStageoutPFN()
79	>	### FEDE added a check for the final /
80	>	if ( endpoint[-1] != '/' ) : endpoint = endpoint + '/'
81	>	######################################
82	>
83	>	#extract SE name an SE_PATH (needed for publication)
84	>	SE, SE_PATH, User = self.splitEndpoint(endpoint)
85	>
86	>	return endpoint, self.lfn , SE, SE_PATH, User
87	>
88	>	def splitEndpoint(self, endpoint):
89	>	'''
90	>	Return relevant infos from endpoint
91	>	'''
92	>	SE = ''
93	>	SE_PATH = ''
94	>	USER = ''
95	>	if self.usePhedex:
96	>	if self.protocol == 'direct':
97	>	query=endpoint
98	>	SE_PATH = endpoint
99	>	SE = self.SE[self.sched]
100	>	else:
101	>	url = 'http://'+endpoint.split('://')[1]
102	>	# python > 2.4
103	>	# SE = urlparse(url).hostname
104	>	scheme, host, path, params, query, fragment = urlparse(url)
105	>	SE = host.split(':')[0]
106	>	SE_PATH = endpoint.split(host)[1]
107	>	USER = (query.split('user')[1]).split('/')[1]
108	>	else:
109	>	#### to test #####
110	>	# url = 'http://'+endpoint.split('://')[1]
111	>	# scheme, host, path, params, query, fragment = urlparse(url)
112	>	# SE = host.split(':')[0]
113	>	# SE_PATH = endpoint.split(host)[1]
114	>	SE = self.node
115	>	SE_PATH = self.user_se_path + self.user_remote_dir
116	>	try:
117	>	USER = (self.lfn.split('user')[1]).split('/')[1]
118	>	except:
119	>	pass
120	>
121	>	return SE, SE_PATH, USER
122	>
123	>
124	>	def getLFN(self):
125	>	"""
126	>	define the LFN composing the needed pieces
127	>	"""
128	>	lfn = ''
129	>	l_User = False
130	>	if not self.usePhedex and (int(self.publish_data) == 0 and int(self.usenamespace) == 0) :
131	>	### add here check if user is trying to force a wrong LFN using a T2  TODO
132	>	## check if storage_name is a T2 (siteDB query)
133	>	## if yes :match self.user_lfn with LFNBaseName...
134	>	##     if NOT : raise (you are using a T2. It's not allowed stage out into self.user_path+self.user_lfn)
135	>	lfn = self.user_remote_dir
136	>	return lfn
137	>	if self.publish_data_name == '' and int(self.publish_data) == 1:
138	>	msg = "Error. The [USER] section does not have 'publish_data_name'\n"
139	>	msg += '\tFor further information please visit : \n\t%s'%self.dataPub_faq
140	>	raise CrabException(msg)
141	>	if self.publish_data_name == '' and int(self.usenamespace) == 1:
142	>	self.publish_data_name = "DefaultDataset"
143	>	if int(self.publish_data) == 1:
144	>	if self.sched in ['CAF']: l_User=True
145	>	primaryDataset = self.computePrimaryDataset()
146	>	### added the case lfn = LFNBase(self.forced_path, primaryDataset, self.publish_data_name, publish=True)
147	>	### for the publication in order to be able to check the lfn length
148	>	lfn = LFNBase(self.forced_path, primaryDataset, self.publish_data_name, publish=True)  + '/${PSETHASH}/'
149	>	elif int(self.usenamespace) == 1:
150	>	if self.sched in ['CAF']: l_User=True
151	>	primaryDataset = self.computePrimaryDataset()
152	>	lfn = LFNBase(self.forced_path, primaryDataset, self.publish_data_name)  + '/${PSETHASH}/'
153	>	else:
154	>	if self.sched in ['CAF','LSF']: l_User=True
155	>	lfn = LFNBase(self.forced_path,self.user_remote_dir)
156	>	return lfn
157	>
158	>	def computePrimaryDataset(self):
159	>	"""
160	>	compute the last part for the LFN in case of publication
161	>	"""
162	>	if (self.datasetpath.upper() != 'NONE'):
163	>	primarydataset = self.datasetpath.split("/")[1]
164	>	else:
165	>	primarydataset = self.publish_data_name
166	>	return primarydataset
167
168	<	def lfn2pfn(self,node,lfn,protocol):
169	<	"""
170	<	PhEDEx Data Service lfn2pfn call
171	<
172	<	input:   LFN,node name,protocol
173	<	returns: DOM object with the content of the PhEDEx Data Service call
174	<	"""
175	<	params = {'node' : node , 'lfn': lfn , 'protocol': protocol}
176	<	params = urllib.urlencode(params)
177	<	#print params
178	<	datasvc_lfn2pfn="%s/lfn2pfn"%self.datasvc_url
179	<	urlresults = urllib.urlopen(datasvc_lfn2pfn, params)
180	<	try:
181	<	urlresults = parse(urlresults)
182	<	except:
183	<	urlresults = None
184	<	return urlresults
185	<
186	<	def parse_error(self,urlresults):
187	<	"""
188	<	look for errors in the DOM object returned by PhEDEx Data Service call
189	<	"""
190	<	errormsg = None
191	<	errors=urlresults.getElementsByTagName('error')
192	<	for error in errors:
193	<	errormsg=error.childNodes[0].data
194	<	if len(error.childNodes)>1:
195	<	errormsg+=error.childNodes[1].data
196	<	return errormsg
197	<
198	<	def parse_lfn2pfn(self,urlresults):
199	<	"""
46	<	Parse the content of the result of lfn2pfn PhEDEx Data Service  call
47	<
48	<	input:    DOM object with the content of the lfn2pfn call
49	<	returns:  PFN
50	<	"""
51	<	result = urlresults.getElementsByTagName('phedex')
52	<	if not result:
53	<	return []
54	<	result = result[0]
55	<	pfn = None
56	<	mapping = result.getElementsByTagName('mapping')
57	<	for m in mapping:
58	<	pfn=m.getAttribute("pfn")
59	<	if pfn:
60	<	return pfn
61	<
62	<	def getStageoutPFN(self,node,lfn,protocol):
63	<	"""
64	<	input:   LFN,node name,protocol
65	<	returns: PFN
66	<	"""
67	<	fullurl="%s/lfn2pfn?node=%s&lfn=%s&protocol=%s"%(self.datasvc_url,node,lfn,protocol)
68	<	domlfn2pfn = self.lfn2pfn(node,lfn,protocol)
69	<	if not domlfn2pfn :
70	<	msg="Unable to get info from %s"%fullurl
71	<	raise CrabException(msg)
72	<
73	<	errormsg = self.parse_error(domlfn2pfn)
74	<	if errormsg:
75	<	msg="Error extracting info from %s due to: %s"%(fullurl,errormsg)
76	<	raise CrabException(msg)
77	<
78	<	stageoutpfn = self.parse_lfn2pfn(domlfn2pfn)
79	<	if not stageoutpfn:
80	<	msg="Unable to get stageout path (PFN) from %s"%fullurl
81	<	raise CrabException(msg)
82	<	return stageoutpfn
83	<
84	<
85	<	if __name__ == '__main__' :
86	<	"""
87	<	"""
88	<	from crab_logger import Logger
89	<	from WorkSpace import *
90	<	continue_dir="/home/fanfani/CRAB"
91	<	cfg_params={'USER.logdir' : continue_dir }
92	<	common.work_space = WorkSpace(continue_dir, cfg_params)
93	<	log = Logger()
94	<	common.logger = log
95	<
96	<	from LFNBaseName import *
97	<	# test values
98	<	lfn = LFNBase("datasetstring")
99	<	node='T2_IT_Bari'
100	<	protocol="srmv2"
101	<
102	<	#create an instance of the PhEDExDatasvcInfo object
103	<	dsvc = PhEDExDatasvcInfo()
104	<	#extract the PFN for the given node,LFN,protocol
105	<	print "Stageout to %s"%dsvc.getStageoutPFN(node,lfn,protocol)
168	>	def lfn2pfn(self):
169	>	"""
170	>	PhEDEx Data Service lfn2pfn call
171	>
172	>	input:   LFN,node name,protocol
173	>	returns: DOM object with the content of the PhEDEx Data Service call
174	>	"""
175	>	params = {'node' : self.node , 'lfn': self.lfn , 'protocol': self.protocol}
176	>	params = urllib.urlencode(params)
177	>	datasvc_lfn2pfn="%s/lfn2pfn"%self.datasvc_url
178	>	try:
179	>	urlresults = urllib.urlopen(datasvc_lfn2pfn, params)
180	>	urlresults = parse(urlresults)
181	>	except IOError:
182	>	msg="Unable to access PhEDEx Data Service at %s"%datasvc_lfn2pfn
183	>	raise CrabException(msg)
184	>	except:
185	>	urlresults = None
186	>
187	>	return urlresults
188	>
189	>	def parse_error(self,urlresults):
190	>	"""
191	>	look for errors in the DOM object returned by PhEDEx Data Service call
192	>	"""
193	>	errormsg = None
194	>	errors=urlresults.getElementsByTagName('error')
195	>	for error in errors:
196	>	errormsg=error.childNodes[0].data
197	>	if len(error.childNodes)>1:
198	>	errormsg+=error.childNodes[1].data
199	>	return errormsg
200
201	+	def parse_lfn2pfn(self,urlresults):
202	+	"""
203	+	Parse the content of the result of lfn2pfn PhEDEx Data Service  call
204	+
205	+	input:    DOM object with the content of the lfn2pfn call
206	+	returns:  PFN
207	+	"""
208	+	result = urlresults.getElementsByTagName('phedex')
209	+
210	+	if not result:
211	+	return []
212	+	result = result[0]
213	+	pfn = None
214	+	mapping = result.getElementsByTagName('mapping')
215	+	for m in mapping:
216	+	pfn=m.getAttribute("pfn")
217	+	if pfn:
218	+	return pfn
219	+
220	+	def getStageoutPFN( self ):
221	+	"""
222	+	input:   LFN,node name,protocol
223	+	returns: PFN
224	+	"""
225	+	if self.usePhedex:
226	+	fullurl="%s/lfn2pfn?node=%s&lfn=%s&protocol=%s"%(self.datasvc_url,self.node,self.lfn,self.protocol)
227	+	domlfn2pfn = self.lfn2pfn()
228	+	if not domlfn2pfn :
229	+	msg="Unable to get info from %s"%fullurl
230	+	raise CrabException(msg)
231	+
232	+	errormsg = self.parse_error(domlfn2pfn)
233	+	if errormsg:
234	+	msg="Error extracting info from %s due to: %s"%(fullurl,errormsg)
235	+	raise CrabException(msg)
236	+
237	+	stageoutpfn = self.parse_lfn2pfn(domlfn2pfn)
238	+	if not stageoutpfn:
239	+	msg ='Unable to get stageout path from TFC at Site %s \n'%self.node
240	+	msg+='      Please alert the CompInfraSup group through their savannah %s \n'%self.FacOps_savannah
241	+	msg+='      reporting: \n'
242	+	msg+='       Summary: Unable to get user stageout from TFC at Site %s \n'%self.node
243	+	msg+='       OriginalSubmission: stageout path is not retrieved from %s \n'%fullurl
244	+	raise CrabException(msg)
245	+	else:
246	+	if self.sched in ['CAF','LSF'] :
247	+	stageoutpfn = self.user_se_path+self.lfn
248	+	else:
249	+	stageoutpfn = 'srm://'+self.node+':'+self.user_port+self.user_se_path+self.lfn
250	+
251	+	return stageoutpfn
252	+
253	+
254	+
255	+	if __name__ == '__main__':
256	+	"""
257	+	Sort of unit testing to check Phedex API for whatever site and/or lfn.
258	+	Usage:
259	+	python PhEDExDatasvcInfo.py --node T2_IT_Bari --lfn /store/maremma
260	+
261	+	"""
262	+	import getopt,sys
263	+	from crab_util import *
264	+	import common
265	+	klass_name = 'SchedulerGlite'
266	+	klass = importName(klass_name, klass_name)
267	+	common.scheduler = klass()
268	+
269	+	lfn="/store/user/"
270	+	node='T2_IT_Bari'
271	+	valid = ['node=','lfn=']
272	+	try:
273	+	opts, args = getopt.getopt(sys.argv[1:], "", valid)
274	+	except getopt.GetoptError, ex:
275	+	print str(ex)
276	+	sys.exit(1)
277	+	for o, a in opts:
278	+	if o == "--node":
279	+	node = a
280	+	if o == "--lfn":
281	+	lfn = a
282	+
283	+	mycfg_params = { 'USER.storage_element': node }
284	+	dsvc = PhEDExDatasvcInfo(mycfg_params)
285	+	dsvc.lfn = lfn
286	+	print dsvc.getStageoutPFN()
287	+

Diff Legend

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