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 1.12 by spiga, Wed Oct 15 13:54:13 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"
16	<	datasvc_url="https://cmsweb.cern.ch/phedex/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_lfn = cfg_params.get("USER.lfn",'')
33	>	if self.user_lfn:
34	>	if ( self.user_lfn[-1] != '/' ) : self.user_lfn = self.user_lfn + '/'
35	>
36	>	self.user_port = cfg_params.get("USER.storage_port",'8443')
37	>	self.user_se_path = cfg_params.get("USER.storage_path",'')
38	>	if self.user_se_path:
39	>	if ( self.user_se_path[-1] != '/' ) : self.user_se_path = self.user_se_path + '/'
40	>
41	>
42	>	#check if using "private" Storage
43	>	self.usePhedex = True
44	>	stage_out_faq='https://twiki.cern.ch/twiki/bin/view/CMS/SWGuideCrabFaq#How_to_store_output_with_CRAB_2'
45	>	if not self.node :
46	>	msg = 'Please specify the storage_element name in your crab.cfg section [USER].\n'
47	>	msg +='      For further information please visit : %s'%stage_out_faq
48	>	raise CrabException(msg)
49	>	if (self.node.find('T1_') + self.node.find('T2_')+self.node.find('T3_')) == -3: self.usePhedex = False
50	>	if not self.usePhedex and ( self.user_lfn == '' or self.user_se_path == '' ):
51	>	msg = 'You are asking to stage out without using CMS Storage Name convention. In this case you \n'
52	>	msg += '      must specify both lfn and storage_path in the crab.cfg section [USER].\n '
53	>	msg += '      For further information please visit : %s'%stage_out_faq
54	>	raise CrabException(msg)
55	>	self.sched = common.scheduler.name().upper()
56	>	self.protocol = self.srm_version
57	>	if self.sched in ['CAF','LSF']:self.protocol = 'direct'
58	>
59	>	return
60	>
61	>	def getEndpoint(self):
62	>	'''
63	>	Return full SE endpoint and related infos
64	>	'''
65	>	self.lfn = self.getLFN()
66	>
67	>	#extract the PFN for the given node,LFN,protocol
68	>	endpoint = self.getStageoutPFN()
69
70	<	def lfn2pfn(self,node,lfn,protocol):
71	<	"""
72	<	PhEDEx Data Service lfn2pfn call
73	<
74	<	input:   LFN,node name,protocol
75	<	returns: DOM object with the content of the PhEDEx Data Service call
76	<	"""
77	<	params = {'node' : node , 'lfn': lfn , 'protocol': protocol}
78	<	params = urllib.urlencode(params)
79	<	#print params
80	<	datasvc_lfn2pfn="%s/lfn2pfn"%self.datasvc_url
81	<	urlresults = urllib.urlopen(datasvc_lfn2pfn, params)
82	<	try:
83	<	urlresults = parse(urlresults)
84	<	except:
85	<	urlresults = None
86	<	return urlresults
87	<
88	<	def parse_error(self,urlresults):
89	<	"""
90	<	look for errors in the DOM object returned by PhEDEx Data Service call
91	<	"""
92	<	errormsg = None
93	<	errors=urlresults.getElementsByTagName('error')
94	<	for error in errors:
95	<	errormsg=error.childNodes[0].data
96	<	if len(error.childNodes)>1:
97	<	errormsg+=error.childNodes[1].data
98	<	return errormsg
99	<
100	<	def parse_lfn2pfn(self,urlresults):
101	<	"""
102	<	Parse the content of the result of lfn2pfn PhEDEx Data Service  call
103	<
104	<	input:    DOM object with the content of the lfn2pfn call
105	<	returns:  PFN
106	<	"""
107	<	result = urlresults.getElementsByTagName('phedex')
108	<	if not result:
109	<	return []
110	<	result = result[0]
111	<	pfn = None
112	<	mapping = result.getElementsByTagName('mapping')
113	<	for m in mapping:
114	<	pfn=m.getAttribute("pfn")
115	<	if pfn:
116	<	return pfn
117	<
118	<	def getStageoutPFN(self,node,lfn,protocol):
119	<	"""
120	<	input:   LFN,node name,protocol
121	<	returns: PFN
122	<	"""
123	<	fullurl="%s/lfn2pfn?node=%s&lfn=%s&protocol=%s"%(self.datasvc_url,node,lfn,protocol)
124	<	domlfn2pfn = self.lfn2pfn(node,lfn,protocol)
125	<	if not domlfn2pfn :
126	<	msg="Unable to get info from %s"%fullurl
127	<	raise CrabException(msg)
128	<
129	<	errormsg = self.parse_error(domlfn2pfn)
130	<	if errormsg:
131	<	msg="Error extracting info from %s due to: %s"%(fullurl,errormsg)
132	<	raise CrabException(msg)
133	<
134	<	stageoutpfn = self.parse_lfn2pfn(domlfn2pfn)
135	<	if not stageoutpfn:
136	<	msg="Unable to get stageout path (PFN) from %s"%fullurl
137	<	raise CrabException(msg)
138	<	return stageoutpfn
139	<
140	<
141	<	if __name__ == '__main__' :
142	<	"""
143	<	"""
144	<	from crab_logger import Logger
145	<	from WorkSpace import *
146	<	continue_dir="/home/fanfani/CRAB"
92	<	cfg_params={'USER.logdir' : continue_dir }
93	<	common.work_space = WorkSpace(continue_dir, cfg_params)
94	<	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)
70	>	#extract SE name an SE_PATH (needed for publication)
71	>	SE, SE_PATH, User = self.splitEndpoint(endpoint)
72	>
73	>	return endpoint, self.lfn , SE, SE_PATH, User
74	>
75	>	def splitEndpoint(self, endpoint):
76	>	'''
77	>	Return relevant infos from endpoint
78	>	'''
79	>	SE = ''
80	>	SE_PATH = ''
81	>	USER = ''
82	>	if self.usePhedex:
83	>	if self.protocol == 'direct':
84	>	query=endpoint
85	>	SE_PATH = endpoint
86	>	### FEDE added SE ###
87	>	SE = self.sched
88	>	else:
89	>	url = 'http://'+endpoint.split('://')[1]
90	>	# python > 2.4
91	>	# SE = urlparse(url).hostname
92	>	scheme, host, path, params, query, fragment = urlparse(url)
93	>	SE = host.split(':')[0]
94	>	SE_PATH = endpoint.split(host)[1]
95	>	USER = (query.split('user')[1]).split('/')[1]
96	>	else:
97	>	SE = self.node
98	>	SE_PATH = self.user_se_path + self.user_lfn
99	>	try:
100	>	USER = (self.lfn.split('user')[1]).split('/')[1]
101	>	except:
102	>	pass
103	>
104	>	return SE, SE_PATH, USER
105	>
106	>
107	>	def getLFN(self):
108	>	"""
109	>	define the LFN composing the needed pieces
110	>	"""
111	>	lfn = ''
112	>	l_User = False
113	>	if not self.usePhedex and (int(self.publish_data) == 0 and int(self.usenamespace) == 0) :
114	>	### add here check if user is trying to force a wrong LFN using a T2  TODO
115	>	## check if storage_name is a T2 (siteDB query)
116	>	## if yes :match self.user_lfn with LFNBaseName...
117	>	##     if NOT : raise (you are using a T2. It's not allowed stage out into self.user_path+self.user_lfn)
118	>	lfn = self.user_lfn
119	>	return lfn
120	>	if self.publish_data_name == '' and int(self.publish_data) == 1:
121	>	msg = "Eeror. The [USER] section does not have 'publish_data_name'"
122	>	raise CrabException(msg)
123	>	if self.publish_data_name == '' and int(self.usenamespace) == 1:
124	>	self.publish_data_name = "DefaultDataset"
125	>	if int(self.publish_data) == 1 or int(self.usenamespace) == 1:
126	>	if self.sched in ['CAF']: l_User=True
127	>	primaryDataset = self.computePrimaryDataset()
128	>	lfn = LFNBase(primaryDataset,self.publish_data_name,LocalUser=l_User)  + '/${PSETHASH}/'
129	>	else:
130	>	if self.sched in ['CAF','LSF']: l_User=True
131	>	lfn = LFNBase(self.user_remote_dir,LocalUser=l_User)
132	>	return lfn
133	>
134	>	def computePrimaryDataset(self):
135	>	"""
136	>	compute the last part for the LFN in case of publication
137	>	"""
138	>	if (self.datasetpath.upper() != 'NONE'):
139	>	primarydataset = self.datasetpath.split("/")[1]
140	>	else:
141	>	primarydataset = self.publish_data_name
142	>	return primarydataset
143	>
144	>	def lfn2pfn(self):
145	>	"""
146	>	PhEDEx Data Service lfn2pfn call
147
148	+	input:   LFN,node name,protocol
149	+	returns: DOM object with the content of the PhEDEx Data Service call
150	+	"""
151	+	params = {'node' : self.node , 'lfn': self.lfn , 'protocol': self.protocol}
152	+	params = urllib.urlencode(params)
153	+	datasvc_lfn2pfn="%s/lfn2pfn"%self.datasvc_url
154	+	urlresults = urllib.urlopen(datasvc_lfn2pfn, params)
155	+	try:
156	+	urlresults = parse(urlresults)
157	+	except:
158	+	urlresults = None
159	+
160	+	return urlresults
161	+
162	+	def parse_error(self,urlresults):
163	+	"""
164	+	look for errors in the DOM object returned by PhEDEx Data Service call
165	+	"""
166	+	errormsg = None
167	+	errors=urlresults.getElementsByTagName('error')
168	+	for error in errors:
169	+	errormsg=error.childNodes[0].data
170	+	if len(error.childNodes)>1:
171	+	errormsg+=error.childNodes[1].data
172	+	return errormsg
173	+
174	+	def parse_lfn2pfn(self,urlresults):
175	+	"""
176	+	Parse the content of the result of lfn2pfn PhEDEx Data Service  call
177	+
178	+	input:    DOM object with the content of the lfn2pfn call
179	+	returns:  PFN
180	+	"""
181	+	result = urlresults.getElementsByTagName('phedex')
182	+
183	+	if not result:
184	+	return []
185	+	result = result[0]
186	+	pfn = None
187	+	mapping = result.getElementsByTagName('mapping')
188	+	for m in mapping:
189	+	pfn=m.getAttribute("pfn")
190	+	if pfn:
191	+	return pfn
192	+
193	+	def getStageoutPFN( self ):
194	+	"""
195	+	input:   LFN,node name,protocol
196	+	returns: PFN
197	+	"""
198	+	if self.usePhedex:
199	+	fullurl="%s/lfn2pfn?node=%s&lfn=%s&protocol=%s"%(self.datasvc_url,self.node,self.lfn,self.protocol)
200	+	domlfn2pfn = self.lfn2pfn()
201	+	if not domlfn2pfn :
202	+	msg="Unable to get info from %s"%fullurl
203	+	raise CrabException(msg)
204	+
205	+	errormsg = self.parse_error(domlfn2pfn)
206	+	if errormsg:
207	+	msg="Error extracting info from %s due to: %s"%(fullurl,errormsg)
208	+	raise CrabException(msg)
209	+
210	+	stageoutpfn = self.parse_lfn2pfn(domlfn2pfn)
211	+	if not stageoutpfn:
212	+	msg ='Unable to get stageout path from TFC at Site %s \n'%self.node
213	+	msg+='      Please alert the CompInfraSup group through their savannah %s \n'%self.FacOps_savannah
214	+	msg+='      reporting: \n'
215	+	msg+='       Summary: Unable to get user stageout from TFC at Site %s \n'%self.node
216	+	msg+='       OriginalSubmission: stageout path is not retrieved from %s \n'%fullurl
217	+	raise CrabException(msg)
218	+	else:
219	+	if self.sched in ['CAF','LSF'] :
220	+	stageoutpfn = self.user_se_path+self.lfn
221	+	else:
222	+	stageoutpfn = 'srm://'+self.node+':'+self.user_port+self.user_se_path+self.lfn
223	+
224	+	return stageoutpfn
225	+
226	+
227	+
228	+	if __name__ == '__main__':
229	+	"""
230	+	Sort of unit testing to check Phedex API for whatever site and/or lfn.
231	+	Usage:
232	+	python PhEDExDatasvcInfo.py --node T2_IT_Bari --lfn /store/maremma
233	+
234	+	"""
235	+	import getopt,sys
236	+	from crab_util import *
237	+	import common
238	+	klass_name = 'SchedulerGlite'
239	+	klass = importName(klass_name, klass_name)
240	+	common.scheduler = klass()
241	+
242	+	lfn="/store/user/"
243	+	node='T2_IT_Bari'
244	+	valid = ['node=','lfn=']
245	+	try:
246	+	opts, args = getopt.getopt(sys.argv[1:], "", valid)
247	+	except getopt.GetoptError, ex:
248	+	print str(ex)
249	+	sys.exit(1)
250	+	for o, a in opts:
251	+	if o == "--node":
252	+	node = a
253	+	if o == "--lfn":
254	+	lfn = a
255	+
256	+	mycfg_params = { 'USER.storage_element': node }
257	+	dsvc = PhEDExDatasvcInfo(mycfg_params)
258	+	dsvc.lfn = lfn
259	+	print dsvc.getStageoutPFN()
260	+

Diff Legend

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