CKAN = CKAN_CLIENT(options.host, options.auth)
## UP = UPLOADER(CKAN, OUT, options.outdir,options.fromdate)
UP = Uploader(CKAN, options.ckan_check, HandleClient, cred, OUT,
options.outdir, options.fromdate, options.host)
if (options.identifier):
list = [options.identifier]
listtext = 'given by option -i (%d id\'s)' % len(list)
elif (options.list):
f = open(options.list, 'r')
list = f.readlines()
f.close()
listtext = 'got from file %s (%d id\'s)' % (options.list, len(list))
elif (options.community):
##UP.purge_group(options.community)
UP.get_packages(options.community)
##HEW??? UP.get_group_list(options.community)
print "--- Start get community list from CKAN---\n"
list = UP.package_list.keys()
##clist = UP.get_packages(options.community).keys()
##print clist
listtext = 'got from CKAN community %s, stored in file %s-id.list (%d id\'s)' % (
options.community, options.community, len(list))
cf = open('%s-id.list' % options.community, 'w')
cf.write("\n".join(list))
cf.close()
##print UP.package_list.keys()
else:
print 'ERROR : one of the otptions -c COMMUNITY, -i IDENTIFIER or -l LIST must be given'
sys.exit()
def process_delete(OUT, dir, options):
print("###JM# Don't use this function. It is not up to date.")
return False
# create CKAN object
CKAN = CKAN_CLIENT(options.iphost, options.auth)
UP = Uploader(CKAN, OUT, options.outdir)
##HEW-D-ec credentials,ec = None,None
# create credentials
try:
cred = b2handle.clientcredentials.PIDClientCredentials.load_from_JSON(
'credentials_11098')
except Exception:
logging.critical(
"[CRITICAL] %s Could not create credentials from credstore %s" %
(options.handle_check))
p.print_help()
sys.exit(-1)
else:
logging.debug(
"Create handle client instance to add uuid to handle server")
for delete_file in glob.glob(dir + '/*.del'):
community, mdprefix = os.path.splitext(
os.path.basename(delete_file))[0].split('-')
logging.info('\n## Deleting datasets from community "%s" ##' %
(community))
# get packages from the group in CKAN:
UP.get_packages(community)
# open the delete file and loop over its lines:
file_content = ''
try:
f = open(delete_file, 'r')
file_content = f.read()
f.close()
except IOError:
logging.critical("Cannot read data from '{0}'".format(delete_file))
f.close
else:
# rename the file in a crash backup file:
os.rename(delete_file, delete_file + '.crash-backup')
results = {'count': 0, 'ecount': 0, 'tcount': 0, 'time': 0}
# use a try-except-finally environment to gurantee that no deleted metadata information will be lost:
try:
logging.info(' | | %-4s | %-50s | %-50s |\n |%s|' %
('#', 'oai identifier', 'CKAN identifier', "-" * 116))
deletestart = time.time()
for line in file_content.split('\n'):
# ignore empty lines:
if not line:
continue
results['tcount'] += 1
subset, identifier = line.split('\t')
# dataset name uniquely generated from oai identifier
uid = uuid.uuid5(uuid.NAMESPACE_DNS,
identifier.encode('ascii', 'replace'))
ds = str(uid)
# output:
logging.info(' | d | %-4d | %-50s | %-50s |' %
(results['tcount'], identifier, ds))
### CHECK STATUS OF DATASET IN CKAN AND PID:
# status of data set
dsstatus = "unknown"
# check against handle server
handlestatus = "unknown"
##HEW-D-ec??? pid = credentials.prefix + "/eudat-jmd_" + ds
pid = "11098/eudat-jmd_" + ds_id
pidRecord["CHECKSUM"] = client.get_value_from_handle(
pid, "CHECKSUM")
if (pidRecord["CHECKSUM"] == None):
logging.debug(
" |-> Can not access pid %s to get checksum" %
(pid))
handlestatus = "new"
else:
logging.debug(" |-> pid %s exists" % (pid))
handlestatus = "exist"
# check against CKAN database
ckanstatus = 'unknown'
ckanstatus = UP.check_dataset(ds, None)
delete = 0
# depending on handle status delete record from B2FIND
if (handlestatus == "new"
and ckanstatus == "new"): # no action required
logging.info(' |-> %s' % ('No deletion required'))
else:
delete = UP.delete(ds, ckanstatus)
if (delete == 1):
logging.info(' |-> %s' %
('Deletion was successful'))
results['count'] += 1
# delete handle (to keep the symmetry between handle and B2FIND server)
if (handlestatus == "exist"):
logging.info(
" |-> Delete handle %s with checksum %s"
% (pid, pidRecord["CHECKSUM"]))
try:
client.delete_handle(pid)
except GenericHandleError as err:
logging.error('[ERROR] Unexpected Error: %s' %
err)
except Exception:
logging.error('[ERROR] Unexpected Error:')
else:
logging.info(
" |-> No action (deletion) required for handle %s"
% pid)
else:
logging.info(' |-> %s' % ('Deletion failed'))
results['ecount'] += 1
except Exception:
logging.error('[ERROR] Unexpected Error')
logging.error('You find the ids of the deleted metadata in "%s"' %
(delete_file + '.crash-backup'))
raise
else:
# all worked fine you can remove the crash-backup file:
os.remove(delete_file + '.crash-backup')
deletetime = time.time() - deletestart
results['time'] = deletetime
# save stats:
OUT.save_stats(community + '-' + mdprefix, subset, 'd', results)
sys.exit(-1)
CKAN = CKAN_CLIENT(options.host,options.auth)
## UP = UPLOADER(CKAN, OUT, options.outdir,options.fromdate)
UP = Uploader(CKAN,options.ckan_check,HandleClient,cred,OUT,options.outdir,options.fromdate,options.host)
if (options.identifier):
list = [ options.identifier ]
listtext='given by option -i (%d id\'s)' % len(list)
elif (options.list):
f = open(options.list,'r')
list = f.readlines()
f.close()
listtext='got from file %s (%d id\'s)' % (options.list,len(list))
elif (options.community):
##UP.purge_group(options.community)
UP.get_packages(options.community)
##HEW??? UP.get_group_list(options.community)
print "--- Start get community list from CKAN---\n"
list = UP.package_list.keys()
##clist = UP.get_packages(options.community).keys()
##print clist
listtext='got from CKAN community %s, stored in file %s-id.list (%d id\'s)' % (options.community,options.community,len(list))
cf = open('%s-id.list' % options.community,'w')
cf.write("\n".join(list))
cf.close()
##print UP.package_list.keys()
else:
print 'ERROR : one of the otptions -c COMMUNITY, -i IDENTIFIER or -l LIST must be given'
sys.exit()
##HEW-Tprint '%s list ' % list
def process_delete(OUT, dir, options):
print ("###JM# Don't use this function. It is not up to date.")
return False
# create CKAN object
CKAN = CKAN_CLIENT(options.iphost,options.auth)
UP = Uploader(CKAN,OUT,options.outdir)
##HEW-D-ec credentials,ec = None,None
# create credentials
try:
cred = b2handle.clientcredentials.PIDClientCredentials.load_from_JSON('credentials_11098')
except Exception:
logging.critical("[CRITICAL] %s Could not create credentials from credstore %s" % (options.handle_check))
p.print_help()
sys.exit(-1)
else:
logging.debug("Create handle client instance to add uuid to handle server")
for delete_file in glob.glob(dir+'/*.del'):
community, mdprefix = os.path.splitext(os.path.basename(delete_file))[0].split('-')
logging.info('\n## Deleting datasets from community "%s" ##' % (community))
# get packages from the group in CKAN:
UP.get_packages(community)
# open the delete file and loop over its lines:
file_content = ''
try:
f = open(delete_file, 'r')
file_content = f.read()
f.close()
except IOError :
logging.critical("Cannot read data from '{0}'".format(delete_file))
f.close
else:
# rename the file in a crash backup file:
os.rename(delete_file,delete_file+'.crash-backup')
results = {
'count':0,
'ecount':0,
'tcount':0,
'time':0
}
# use a try-except-finally environment to gurantee that no deleted metadata information will be lost:
try:
logging.info(' | | %-4s | %-50s | %-50s |\n |%s|' % ('#','oai identifier','CKAN identifier',"-" * 116))
deletestart = time.time()
for line in file_content.split('\n'):
# ignore empty lines:
if not line:
continue
results['tcount'] += 1
subset, identifier = line.split('\t')
# dataset name uniquely generated from oai identifier
uid = uuid.uuid5(uuid.NAMESPACE_DNS, identifier.encode('ascii','replace'))
ds = str(uid)
# output:
logging.info(' | d | %-4d | %-50s | %-50s |' % (results['tcount'],identifier,ds))
### CHECK STATUS OF DATASET IN CKAN AND PID:
# status of data set
dsstatus="unknown"
# check against handle server
handlestatus="unknown"
##HEW-D-ec??? pid = credentials.prefix + "/eudat-jmd_" + ds
pid = "11098/eudat-jmd_" + ds_id
pidRecord["CHECKSUM"] = client.get_value_from_handle(pid, "CHECKSUM")
if (pidRecord["CHECKSUM"] == None):
logging.debug(" |-> Can not access pid %s to get checksum" % (pid))
handlestatus="new"
else:
logging.debug(" |-> pid %s exists" % (pid))
handlestatus="exist"
# check against CKAN database
ckanstatus = 'unknown'
ckanstatus=UP.check_dataset(ds,None)
delete = 0
# depending on handle status delete record from B2FIND
if ( handlestatus == "new" and ckanstatus == "new") : # no action required
logging.info(' |-> %s' % ('No deletion required'))
else:
delete = UP.delete(ds,ckanstatus)
if (delete == 1):
logging.info(' |-> %s' % ('Deletion was successful'))
results['count'] += 1
# delete handle (to keep the symmetry between handle and B2FIND server)
if (handlestatus == "exist"):
logging.info(" |-> Delete handle %s with checksum %s" % (pid,pidRecord["CHECKSUM"]))
try:
client.delete_handle(pid)
except GenericHandleError as err:
logging.error('[ERROR] Unexpected Error: %s' % err)
except Exception:
logging.error('[ERROR] Unexpected Error:')
else:
logging.info(" |-> No action (deletion) required for handle %s" % pid)
else:
logging.info(' |-> %s' % ('Deletion failed'))
results['ecount'] += 1
except Exception:
logging.error('[ERROR] Unexpected Error')
logging.error('You find the ids of the deleted metadata in "%s"' % (delete_file+'.crash-backup'))
raise
else:
# all worked fine you can remove the crash-backup file:
os.remove(delete_file+'.crash-backup')
deletetime=time.time()-deletestart
results['time'] = deletetime
# save stats:
OUT.save_stats(community+'-'+mdprefix,subset,'d',results)