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 1.15 by spiga, Sat Nov 8 10:13:25 2008 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 crab_logger import Logger
6	+	from WorkSpace import *
7	+	from urlparse import urlparse
8	+	from LFNBaseName import *
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"
11	>	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	>	self.FacOps_savannah = 'https://savannah.cern.ch/support/?func=additem&group=cmscompinfrasup'
18	>
19	>
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	>	self.user_remote_dir = cfg_params.get("USER.user_remote_dir",'')
26	>	if self.user_remote_dir:
27	>	if ( self.user_remote_dir[-1] != '/' ) : self.user_remote_dir = self.user_remote_dir + '/'
28	>
29	>	self.datasetpath = cfg_params.get("CMSSW.datasetpath")
30	>	self.publish_data_name = cfg_params.get('USER.publish_data_name','')
31	>
32	>	self.user_port = cfg_params.get("USER.storage_port",'8443')
33	>	self.user_se_path = cfg_params.get("USER.storage_path",'')
34	>	if self.user_se_path:
35	>	if ( self.user_se_path[-1] != '/' ) : self.user_se_path = self.user_se_path + '/'
36	>
37	>	#check if using "private" Storage
38	>	self.usePhedex = True
39	>	stage_out_faq='https://twiki.cern.ch/twiki/bin/view/CMS/SWGuideCrabFaq#How_to_store_output_with_CRAB_2'
40	>	if not self.node :
41	>	msg = 'Please specify the storage_element name in your crab.cfg section [USER].\n'
42	>	msg +='      For further information please visit : %s'%stage_out_faq
43	>	raise CrabException(msg)
44	>	if (self.node.find('T1_') + self.node.find('T2_')+self.node.find('T3_')) == -3: self.usePhedex = False
45	>	if not self.usePhedex :
46	>	self.user_lfn = cfg_params.get("USER.lfn",'')
47	>	if self.user_lfn:
48	>	if ( self.user_lfn[-1] != '/' ) : self.user_lfn = self.user_lfn + '/'
49	>	if ( self.user_lfn[1] != '/' ) : self.user_lfn = '/' + self.user_lfn
50	>	if self.user_remote_dir:
51	>	if ( self.user_remote_dir[-1] != '/' ) : self.user_remote_dir = self.user_remote_dir + '/'
52	>	if ( self.user_remote_dir[1] != '/' ) : self.user_remote_dir = '/' + self.user_remote_dir
53	>	if self.user_remote_dir == self.user_lfn :
54	>	msg =  'WARINING: In your crab.cfg you are specifying both lfn and user_remote_dir parameters.\n'
55	>	msg += '\t lfn is now deprecated. Please use only user_remote_dir removing lfn from your crab.cfg'
56	>	common.logger.message(msg)
57	>	else:
58	>	msg =  'ERROR: In your crab.cfg you are specifying both lfn and user_remote_dir parameters giving different values. \n'
59	>	msg += '\t lfn is now deprecated. Please use only user_remote_dir removing lfn from your crab.cfg'
60	>	raise CrabException(msg)
61	>	else:
62	>	msg =  'WARINING: In your crab.cfg you are using "lfn" which is now a deprecating parameter.\n'
63	>	msg += '\t Please use user_remote_dir removing lfn from your crab.cfg'
64	>	common.logger.message(msg)
65	>	self.user_remote_dir = self.user_lfn
66	>
67	>	if not self.usePhedex and ( self.user_remote_dir == '' or self.user_se_path == '' ):
68	>	msg = 'You are asking to stage out without using CMS Storage Name convention. In this case you \n'
69	>	msg += '\t must specify both user_remote_dir and storage_path in the crab.cfg section [USER].\n '
70	>	msg += '\t For further information please visit : %s'%stage_out_faq
71	>	raise CrabException(msg)
72	>	self.sched = common.scheduler.name().upper()
73	>	self.protocol = self.srm_version
74	>	if self.sched in ['CAF','LSF']:self.protocol = 'direct'
75	>
76	>	self.forced_path = '/store/user/'
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	>
90	>	#extract SE name an SE_PATH (needed for publication)
91	>	SE, SE_PATH, User = self.splitEndpoint(endpoint)
92	>
93	>	return endpoint, self.lfn , SE, SE_PATH, User
94	>
95	>	def splitEndpoint(self, endpoint):
96	>	'''
97	>	Return relevant infos from endpoint
98	>	'''
99	>	SE = ''
100	>	SE_PATH = ''
101	>	USER = ''
102	>	if self.usePhedex:
103	>	if self.protocol == 'direct':
104	>	query=endpoint
105	>	SE_PATH = endpoint
106	>	SE = self.sched
107	>	else:
108	>	url = 'http://'+endpoint.split('://')[1]
109	>	# python > 2.4
110	>	# SE = urlparse(url).hostname
111	>	scheme, host, path, params, query, fragment = urlparse(url)
112	>	SE = host.split(':')[0]
113	>	SE_PATH = endpoint.split(host)[1]
114	>	USER = (query.split('user')[1]).split('/')[1]
115	>	else:
116	>	SE = self.node
117	>	SE_PATH = self.user_se_path + self.user_remote_dir
118	>	try:
119	>	USER = (self.lfn.split('user')[1]).split('/')[1]
120	>	except:
121	>	pass
122	>
123	>	return SE, SE_PATH, USER
124	>
125	>
126	>	def getLFN(self):
127	>	"""
128	>	define the LFN composing the needed pieces
129	>	"""
130	>	lfn = ''
131	>	l_User = False
132	>	if not self.usePhedex and (int(self.publish_data) == 0 and int(self.usenamespace) == 0) :
133	>	### add here check if user is trying to force a wrong LFN using a T2  TODO
134	>	## check if storage_name is a T2 (siteDB query)
135	>	## if yes :match self.user_lfn with LFNBaseName...
136	>	##     if NOT : raise (you are using a T2. It's not allowed stage out into self.user_path+self.user_lfn)
137	>	lfn = self.user_remote_dir
138	>	return lfn
139	>	if self.publish_data_name == '' and int(self.publish_data) == 1:
140	>	msg = "Eeror. The [USER] section does not have 'publish_data_name'"
141	>	raise CrabException(msg)
142	>	if self.publish_data_name == '' and int(self.usenamespace) == 1:
143	>	self.publish_data_name = "DefaultDataset"
144	>	if int(self.publish_data) == 1 or int(self.usenamespace) == 1:
145	>	if self.sched in ['CAF']: l_User=True
146	>	primaryDataset = self.computePrimaryDataset()
147	>	lfn = LFNBase(self.forced_path, primaryDataset, self.publish_data_name, LocalUser=l_User)  + '/${PSETHASH}/'
148	>	else:
149	>	if self.sched in ['CAF','LSF']: l_User=True
150	>	lfn = LFNBase(self.forced_path,self.user_remote_dir,LocalUser=l_User)
151	>	return lfn
152	>
153	>	def computePrimaryDataset(self):
154	>	"""
155	>	compute the last part for the LFN in case of publication
156	>	"""
157	>	if (self.datasetpath.upper() != 'NONE'):
158	>	primarydataset = self.datasetpath.split("/")[1]
159	>	else:
160	>	primarydataset = self.publish_data_name
161	>	return primarydataset
162
163	<	def lfn2pfn(self,node,lfn,protocol):
164	<	"""
165	<	PhEDEx Data Service lfn2pfn call
166	<
167	<	input:   LFN,node name,protocol
168	<	returns: DOM object with the content of the PhEDEx Data Service call
169	<	"""
170	<	params = {'node' : node , 'lfn': lfn , 'protocol': protocol}
171	<	params = urllib.urlencode(params)
172	<	#print params
173	<	datasvc_lfn2pfn="%s/lfn2pfn"%self.datasvc_url
174	<	urlresults = urllib.urlopen(datasvc_lfn2pfn, params)
175	<	try:
176	<	urlresults = parse(urlresults)
177	<	except:
178	<	urlresults = None
179	<	return urlresults
180	<
181	<	def parse_error(self,urlresults):
182	<	"""
183	<	look for errors in the DOM object returned by PhEDEx Data Service call
184	<	"""
185	<	errormsg = None
186	<	errors=urlresults.getElementsByTagName('error')
187	<	for error in errors:
188	<	errormsg=error.childNodes[0].data
189	<	if len(error.childNodes)>1:
190	<	errormsg+=error.childNodes[1].data
191	<	return errormsg
43	<
44	<	def parse_lfn2pfn(self,urlresults):
45	<	"""
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)
163	>	def lfn2pfn(self):
164	>	"""
165	>	PhEDEx Data Service lfn2pfn call
166	>
167	>	input:   LFN,node name,protocol
168	>	returns: DOM object with the content of the PhEDEx Data Service call
169	>	"""
170	>	params = {'node' : self.node , 'lfn': self.lfn , 'protocol': self.protocol}
171	>	params = urllib.urlencode(params)
172	>	datasvc_lfn2pfn="%s/lfn2pfn"%self.datasvc_url
173	>	urlresults = urllib.urlopen(datasvc_lfn2pfn, params)
174	>	try:
175	>	urlresults = parse(urlresults)
176	>	except:
177	>	urlresults = None
178	>
179	>	return urlresults
180	>
181	>	def parse_error(self,urlresults):
182	>	"""
183	>	look for errors in the DOM object returned by PhEDEx Data Service call
184	>	"""
185	>	errormsg = None
186	>	errors=urlresults.getElementsByTagName('error')
187	>	for error in errors:
188	>	errormsg=error.childNodes[0].data
189	>	if len(error.childNodes)>1:
190	>	errormsg+=error.childNodes[1].data
191	>	return errormsg
192
193	+	def parse_lfn2pfn(self,urlresults):
194	+	"""
195	+	Parse the content of the result of lfn2pfn PhEDEx Data Service  call
196	+
197	+	input:    DOM object with the content of the lfn2pfn call
198	+	returns:  PFN
199	+	"""
200	+	result = urlresults.getElementsByTagName('phedex')
201	+
202	+	if not result:
203	+	return []
204	+	result = result[0]
205	+	pfn = None
206	+	mapping = result.getElementsByTagName('mapping')
207	+	for m in mapping:
208	+	pfn=m.getAttribute("pfn")
209	+	if pfn:
210	+	return pfn
211	+
212	+	def getStageoutPFN( self ):
213	+	"""
214	+	input:   LFN,node name,protocol
215	+	returns: PFN
216	+	"""
217	+	if self.usePhedex:
218	+	fullurl="%s/lfn2pfn?node=%s&lfn=%s&protocol=%s"%(self.datasvc_url,self.node,self.lfn,self.protocol)
219	+	domlfn2pfn = self.lfn2pfn()
220	+	if not domlfn2pfn :
221	+	msg="Unable to get info from %s"%fullurl
222	+	raise CrabException(msg)
223	+
224	+	errormsg = self.parse_error(domlfn2pfn)
225	+	if errormsg:
226	+	msg="Error extracting info from %s due to: %s"%(fullurl,errormsg)
227	+	raise CrabException(msg)
228	+
229	+	stageoutpfn = self.parse_lfn2pfn(domlfn2pfn)
230	+	if not stageoutpfn:
231	+	msg ='Unable to get stageout path from TFC at Site %s \n'%self.node
232	+	msg+='      Please alert the CompInfraSup group through their savannah %s \n'%self.FacOps_savannah
233	+	msg+='      reporting: \n'
234	+	msg+='       Summary: Unable to get user stageout from TFC at Site %s \n'%self.node
235	+	msg+='       OriginalSubmission: stageout path is not retrieved from %s \n'%fullurl
236	+	raise CrabException(msg)
237	+	else:
238	+	if self.sched in ['CAF','LSF'] :
239	+	stageoutpfn = self.user_se_path+self.lfn
240	+	else:
241	+	stageoutpfn = 'srm://'+self.node+':'+self.user_port+self.user_se_path+self.lfn
242	+
243	+	return stageoutpfn
244	+
245	+
246	+
247	+	if __name__ == '__main__':
248	+	"""
249	+	Sort of unit testing to check Phedex API for whatever site and/or lfn.
250	+	Usage:
251	+	python PhEDExDatasvcInfo.py --node T2_IT_Bari --lfn /store/maremma
252	+
253	+	"""
254	+	import getopt,sys
255	+	from crab_util import *
256	+	import common
257	+	klass_name = 'SchedulerGlite'
258	+	klass = importName(klass_name, klass_name)
259	+	common.scheduler = klass()
260	+
261	+	lfn="/store/user/"
262	+	node='T2_IT_Bari'
263	+	valid = ['node=','lfn=']
264	+	try:
265	+	opts, args = getopt.getopt(sys.argv[1:], "", valid)
266	+	except getopt.GetoptError, ex:
267	+	print str(ex)
268	+	sys.exit(1)
269	+	for o, a in opts:
270	+	if o == "--node":
271	+	node = a
272	+	if o == "--lfn":
273	+	lfn = a
274	+
275	+	mycfg_params = { 'USER.storage_element': node }
276	+	dsvc = PhEDExDatasvcInfo(mycfg_params)
277	+	dsvc.lfn = lfn
278	+	print dsvc.getStageoutPFN()
279	+

Diff Legend

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