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 1.26.2.3 by spiga, Sat Sep 19 11:15:57 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=None, config=None ):
11	>
12	>	## PhEDEx Data Service URL
13	>	self.datasvc_url="https://cmsweb.cern.ch/phedex/datasvc/xml/prod"
14	>
15	>	self.FacOps_savannah = 'https://savannah.cern.ch/support/?func=additem&group=cmscompinfrasup'
16	>	stage_out_faq='https://twiki.cern.ch/twiki/bin/view/CMS/SWGuideCrabHowTo#Stageout_and_publication'
17	>	self.dataPub_faq = 'https://twiki.cern.ch/twiki/bin/view/CMS/SWGuideCrabForPublication'
18	>
19	>	self.usePhedex = True
20	>	self.sched = common.scheduler.name().upper()
21	>
22	>	if config!=None:
23	>	self.checkConfig(config)
24	>	else:
25	>	self.checkCfgConfig(cfg_params)
26	>
27	>	self.protocol = self.srm_version
28	>
29	>
30	>	def checkConfig(self,config):
31	>	"""
32	>	"""
33	>	self.srm_version = config.get("srm_version",'srmv2')
34	>	self.node = config.get('storage_element',None)
35	>	self.lfn='/store/'
36	>
37	>	def checkCfgConfig(self,cfg_params):
38	>	"""
39	>	"""
40	>	self.datasvc_url = cfg_params.get("USER.datasvc_url",self.datasvc_url)
41	>	self.srm_version = cfg_params.get("USER.srm_version",'srmv2')
42	>	self.node = cfg_params.get('USER.storage_element',None)
43	>
44	>	self.publish_data = cfg_params.get("USER.publish_data",0)
45	>	self.usenamespace = cfg_params.get("USER.usenamespace",0)
46	>	self.user_remote_dir = cfg_params.get("USER.user_remote_dir",'')
47	>	if self.user_remote_dir:
48	>	if ( self.user_remote_dir[-1] != '/' ) : self.user_remote_dir = self.user_remote_dir + '/'
49	>
50	>	self.datasetpath = cfg_params.get("CMSSW.datasetpath")
51	>	self.publish_data_name = cfg_params.get('USER.publish_data_name','')
52	>
53	>	self.user_port = cfg_params.get("USER.storage_port",'8443')
54	>	self.user_se_path = cfg_params.get("USER.storage_path",'')
55	>	if self.user_se_path:
56	>	if ( self.user_se_path[-1] != '/' ) : self.user_se_path = self.user_se_path + '/'
57	>
58	>	#check if using "private" Storage
59	>	if not self.node :
60	>	msg = 'Please specify the storage_element name in your crab.cfg section [USER].\n'
61	>	msg +='\tFor further information please visit : %s'%stage_out_faq
62	>	raise CrabException(msg)
63	>	if (self.node.find('T1_') + self.node.find('T2_')+self.node.find('T3_')) == -3: self.usePhedex = False
64	>
65	>	if not self.usePhedex and ( self.user_remote_dir == '' or self.user_se_path == '' ):
66	>	msg = 'You are asking to stage out without using CMS Storage Name convention. In this case you \n'
67	>	msg += '\t must specify both user_remote_dir and storage_path in the crab.cfg section [USER].\n '
68	>	msg += '\t For further information please visit : \n\t%s'%stage_out_faq
69	>	raise CrabException(msg)
70	>
71	>	self.forced_path = '/store/user/'
72	>	if self.sched in ['CAF','LSF']:
73	>	self.protocol = 'direct'
74	>	self.SE = {'CAF':'caf.cern.ch', 'LSF':''}
75	>	if self.sched == 'CAF': self.forced_path = '/store/caf/user/'
76	>
77	>	if not self.usePhedex:
78	>	self.forced_path = self.user_remote_dir
79	>	return
80	>
81	>	def getEndpoint(self):
82	>	'''
83	>	Return full SE endpoint and related infos
84	>	'''
85	>	self.lfn = self.getLFN()
86	>
87	>	#extract the PFN for the given node,LFN,protocol
88	>	endpoint = self.getStageoutPFN()
89	>	### FEDE added a check for the final /
90	>	if ( endpoint[-1] != '/' ) : endpoint = endpoint + '/'
91	>	######################################
92	>
93	>	#extract SE name an SE_PATH (needed for publication)
94	>	SE, SE_PATH, User = self.splitEndpoint(endpoint)
95	>
96	>	return endpoint, self.lfn , SE, SE_PATH, User
97	>
98	>	def splitEndpoint(self, endpoint):
99	>	'''
100	>	Return relevant infos from endpoint
101	>	'''
102	>	SE = ''
103	>	SE_PATH = ''
104	>	USER = ''
105	>	if self.usePhedex:
106	>	if self.protocol == 'direct':
107	>	query=endpoint
108	>	SE_PATH = endpoint
109	>	SE = self.SE[self.sched]
110	>	else:
111	>	url = 'http://'+endpoint.split('://')[1]
112	>	# python > 2.4
113	>	# SE = urlparse(url).hostname
114	>	scheme, host, path, params, query, fragment = urlparse(url)
115	>	SE = host.split(':')[0]
116	>	SE_PATH = endpoint.split(host)[1]
117	>	USER = (query.split('user')[1]).split('/')[1]
118	>	else:
119	>	#### to test #####
120	>	# url = 'http://'+endpoint.split('://')[1]
121	>	# scheme, host, path, params, query, fragment = urlparse(url)
122	>	# SE = host.split(':')[0]
123	>	# SE_PATH = endpoint.split(host)[1]
124	>	SE = self.node
125	>	SE_PATH = self.user_se_path + self.user_remote_dir
126	>	try:
127	>	USER = (self.lfn.split('user')[1]).split('/')[1]
128	>	except:
129	>	pass
130	>
131	>	return SE, SE_PATH, USER
132	>
133	>
134	>	def getLFN(self):
135	>	"""
136	>	define the LFN composing the needed pieces
137	>	"""
138	>	lfn = ''
139	>	l_User = False
140	>	if not self.usePhedex and (int(self.publish_data) == 0 and int(self.usenamespace) == 0) :
141	>	### add here check if user is trying to force a wrong LFN using a T2  TODO
142	>	## check if storage_name is a T2 (siteDB query)
143	>	## if yes :match self.user_lfn with LFNBaseName...
144	>	##     if NOT : raise (you are using a T2. It's not allowed stage out into self.user_path+self.user_lfn)
145	>	lfn = self.user_remote_dir
146	>	return lfn
147	>	if self.publish_data_name == '' and int(self.publish_data) == 1:
148	>	msg = "Error. The [USER] section does not have 'publish_data_name'\n"
149	>	msg += '\tFor further information please visit : \n\t%s'%self.dataPub_faq
150	>	raise CrabException(msg)
151	>	if self.publish_data_name == '' and int(self.usenamespace) == 1:
152	>	self.publish_data_name = "DefaultDataset"
153	>	if int(self.publish_data) == 1:
154	>	if self.sched in ['CAF']: l_User=True
155	>	primaryDataset = self.computePrimaryDataset()
156	>	### added the case lfn = LFNBase(self.forced_path, primaryDataset, self.publish_data_name, publish=True)
157	>	### for the publication in order to be able to check the lfn length
158	>	lfn = LFNBase(self.forced_path, primaryDataset, self.publish_data_name, publish=True)  + '/${PSETHASH}/'
159	>	elif int(self.usenamespace) == 1:
160	>	if self.sched in ['CAF']: l_User=True
161	>	primaryDataset = self.computePrimaryDataset()
162	>	lfn = LFNBase(self.forced_path, primaryDataset, self.publish_data_name)  + '/${PSETHASH}/'
163	>	else:
164	>	if self.sched in ['CAF','LSF']: l_User=True
165	>	lfn = LFNBase(self.forced_path,self.user_remote_dir)
166	>	return lfn
167	>
168	>	def computePrimaryDataset(self):
169	>	"""
170	>	compute the last part for the LFN in case of publication
171	>	"""
172	>	if (self.datasetpath.upper() != 'NONE'):
173	>	primarydataset = self.datasetpath.split("/")[1]
174	>	else:
175	>	primarydataset = self.publish_data_name
176	>	return primarydataset
177
178	<	def lfn2pfn(self,node,lfn,protocol):
179	<	"""
180	<	PhEDEx Data Service lfn2pfn call
181	<
182	<	input:   LFN,node name,protocol
183	<	returns: DOM object with the content of the PhEDEx Data Service call
184	<	"""
185	<	params = {'node' : node , 'lfn': lfn , 'protocol': protocol}
186	<	params = urllib.urlencode(params)
187	<	#print params
188	<	datasvc_lfn2pfn="%s/lfn2pfn"%self.datasvc_url
189	<	urlresults = urllib.urlopen(datasvc_lfn2pfn, params)
190	<	try:
191	<	urlresults = parse(urlresults)
192	<	except:
193	<	urlresults = None
194	<	return urlresults
195	<
196	<	def parse_error(self,urlresults):
197	<	"""
198	<	look for errors in the DOM object returned by PhEDEx Data Service call
199	<	"""
200	<	errormsg = None
201	<	errors=urlresults.getElementsByTagName('error')
202	<	for error in errors:
203	<	errormsg=error.childNodes[0].data
204	<	if len(error.childNodes)>1:
205	<	errormsg+=error.childNodes[1].data
206	<	return errormsg
207	<
208	<	def parse_lfn2pfn(self,urlresults):
209	<	"""
210	<	Parse the content of the result of lfn2pfn PhEDEx Data Service  call
211	<
212	<	input:    DOM object with the content of the lfn2pfn call
213	<	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)
178	>	def lfn2pfn(self):
179	>	"""
180	>	PhEDEx Data Service lfn2pfn call
181	>
182	>	input:   LFN,node name,protocol
183	>	returns: DOM object with the content of the PhEDEx Data Service call
184	>	"""
185	>	params = {'node' : self.node , 'lfn': self.lfn , 'protocol': self.protocol}
186	>	params = urllib.urlencode(params)
187	>	datasvc_lfn2pfn="%s/lfn2pfn"%self.datasvc_url
188	>	try:
189	>	urlresults = urllib.urlopen(datasvc_lfn2pfn, params)
190	>	urlresults = parse(urlresults)
191	>	except IOError:
192	>	msg="Unable to access PhEDEx Data Service at %s"%datasvc_lfn2pfn
193	>	raise CrabException(msg)
194	>	except:
195	>	urlresults = None
196	>
197	>	return urlresults
198	>
199	>	def parse_error(self,urlresults):
200	>	"""
201	>	look for errors in the DOM object returned by PhEDEx Data Service call
202	>	"""
203	>	errormsg = None
204	>	errors=urlresults.getElementsByTagName('error')
205	>	for error in errors:
206	>	errormsg=error.childNodes[0].data
207	>	if len(error.childNodes)>1:
208	>	errormsg+=error.childNodes[1].data
209	>	return errormsg
210	>
211	>	def parse_lfn2pfn(self,urlresults):
212	>	"""
213	>	Parse the content of the result of lfn2pfn PhEDEx Data Service  call
214
215	+	input:    DOM object with the content of the lfn2pfn call
216	+	returns:  PFN
217	+	"""
218	+	result = urlresults.getElementsByTagName('phedex')
219	+
220	+	if not result:
221	+	return []
222	+	result = result[0]
223	+	pfn = None
224	+	mapping = result.getElementsByTagName('mapping')
225	+	for m in mapping:
226	+	pfn=m.getAttribute("pfn")
227	+	if pfn:
228	+	return pfn
229	+
230	+	def getStageoutPFN( self ):
231	+	"""
232	+	input:   LFN,node name,protocol
233	+	returns: PFN
234	+	"""
235	+	if self.usePhedex:
236	+	fullurl="%s/lfn2pfn?node=%s&lfn=%s&protocol=%s"%(self.datasvc_url,self.node,self.lfn,self.protocol)
237	+	domlfn2pfn = self.lfn2pfn()
238	+	if not domlfn2pfn :
239	+	msg="Unable to get info from %s"%fullurl
240	+	raise CrabException(msg)
241	+
242	+	errormsg = self.parse_error(domlfn2pfn)
243	+	if errormsg:
244	+	msg="Error extracting info from %s due to: %s"%(fullurl,errormsg)
245	+	raise CrabException(msg)
246	+
247	+	stageoutpfn = self.parse_lfn2pfn(domlfn2pfn)
248	+	if not stageoutpfn:
249	+	msg ='Unable to get stageout path from TFC at Site %s \n'%self.node
250	+	msg+='      Please alert the CompInfraSup group through their savannah %s \n'%self.FacOps_savannah
251	+	msg+='      reporting: \n'
252	+	msg+='       Summary: Unable to get user stageout from TFC at Site %s \n'%self.node
253	+	msg+='       OriginalSubmission: stageout path is not retrieved from %s \n'%fullurl
254	+	raise CrabException(msg)
255	+	else:
256	+	if self.sched in ['CAF','LSF'] :
257	+	stageoutpfn = self.user_se_path+self.lfn
258	+	else:
259	+	stageoutpfn = 'srm://'+self.node+':'+self.user_port+self.user_se_path+self.lfn
260	+
261	+	return stageoutpfn
262	+
263	+
264	+
265	+	if __name__ == '__main__':
266	+	"""
267	+	Sort of unit testing to check Phedex API for whatever site and/or lfn.
268	+	Usage:
269	+	python PhEDExDatasvcInfo.py --node T2_IT_Bari --lfn /store/maremma
270	+
271	+	"""
272	+	import getopt,sys
273	+	from crab_util import *
274	+	import common
275	+	klass_name = 'SchedulerGlite'
276	+	klass = importName(klass_name, klass_name)
277	+	common.scheduler = klass()
278	+
279	+	lfn="/store/user/"
280	+	node='T2_IT_Bari'
281	+	valid = ['node=','lfn=']
282	+	try:
283	+	opts, args = getopt.getopt(sys.argv[1:], "", valid)
284	+	except getopt.GetoptError, ex:
285	+	print str(ex)
286	+	sys.exit(1)
287	+	for o, a in opts:
288	+	if o == "--node":
289	+	node = a
290	+	if o == "--lfn":
291	+	lfn = a
292	+
293	+	mycfg_params = { 'USER.storage_element': node }
294	+	dsvc = PhEDExDatasvcInfo(mycfg_params)
295	+	dsvc.lfn = lfn
296	+	print dsvc.getStageoutPFN()
297	+

Diff Legend

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