def testMigrateStoreWithSpaces(self):
dest = Location.get_location('test')
local = Location.get_location('local')
datafile, replica = generate_datafile('1/1/Hi Mum', self.dataset,
"Hi mum")
datafile2, replica2 = generate_datafile('1/1/Hi Dad', self.dataset,
"Hi dad")
path = datafile.get_absolute_filepath()
self.assertTrue(os.path.exists(path))
path2 = datafile.get_absolute_filepath()
self.assertTrue(os.path.exists(path2))
# Migrate them
migrate_replica(replica, dest)
self.assertFalse(os.path.exists(path))
migrate_replica(replica2, dest)
self.assertFalse(os.path.exists(path2))
# Bring them back
migrate_replica(datafile.get_preferred_replica(), local)
self.assertTrue(os.path.exists(path))
migrate_replica(datafile2.get_preferred_replica(), local)
self.assertTrue(os.path.exists(path2))
def setUp(self):
self.user = generate_user('fred')
Location.force_initialize()
self.experiment = generate_experiment(users=[self.user])
self.dataset = generate_dataset(experiments=[self.experiment])
self.server = SimpleHttpTestServer()
self.server.start()
def testReplicaVerify(self):
from django.conf import settings
saved = settings.REQUIRE_DATAFILE_CHECKSUMS
try:
Location.get_location('test')
datafile, replica = generate_datafile("1/2/3", self.dataset,
"Hi mum")
settings.REQUIRE_DATAFILE_CHECKSUMS = True
self.assertTrue(replica.verify(), 'Replica.verify() failed.')
replica.datafile.sha512sum = ''
replica.datafile.md5sum = ''
self.assertFalse(
replica.verify(),
'Replica.verify() succeeded despite no checksum '
'(settings.REQUIRE_DATAFILE_CHECKSUMS=True).')
self.assertFalse(
replica.verify(allowEmptyChecksums=False),
'Replica.verify() succeeded despite no checksum '
'(allowEmptyChecksums=False)')
settings.REQUIRE_DATAFILE_CHECKSUMS = False
datafile.sha512sum = None
datafile.md5sum = None
self.assertTrue(
replica.verify(allowEmptyChecksums=True),
'Replica.verify() failed wrongly '
'(allowEmptyChecksums=True)')
datafile.sha512sum = None
datafile.md5sum = None
self.assertTrue(replica.verify(),
'Replica.verify() failed wrongly')
finally:
settings.REQUIRE_DATAFILE_CHECKSUMS = saved
def test_location(self):
from tardis.tardis_portal.models import Location
self.assertEquals(Location.get_default_location().name,
'local')
self.assertEquals(Location.get_location('staging').name,
'staging')
self.assertEquals(len(Location.objects.all()), 6)
def test_location(self):
from tardis.tardis_portal.models import Location
self.assertEquals(Location.get_default_location().name,
'local')
self.assertEquals(Location.get_location('staging').name,
'staging')
self.assertEquals(len(Location.objects.all()), 6)
def setUp(self):
from tardis.tardis_portal import models
from tempfile import mkdtemp, mktemp
from django.conf import settings
from os import path
import os
# Disconnect post_save signal
from django.db.models.signals import post_save
from tardis.tardis_portal.models import \
staging_hook, Dataset_File, Replica, Location
post_save.disconnect(staging_hook, sender=Replica)
from django.contrib.auth.models import User
user = '******'
pwd = 'secret'
email = ''
self.user = User.objects.create_user(user, email, pwd)
try:
os.makedirs(settings.GET_FULL_STAGING_PATH_TEST)
except OSError:
pass
self.temp = mkdtemp(dir=settings.GET_FULL_STAGING_PATH_TEST)
self.file = mktemp(dir=self.temp)
content = 'test file'
with open(self.file, "w+b") as f:
f.write(content)
Location.force_initialize()
# make datafile
exp = models.Experiment(title='test exp1',
institution_name='monash',
created_by=self.user)
exp.save()
# make dataset
dataset = models.Dataset(description="dataset description...")
dataset.save()
dataset.experiments.add(exp)
dataset.save()
# create datafile
df = models.Dataset_File(dataset=dataset, size = len(content),
filename = path.basename(self.file),
md5sum='f20d9f2072bbeb6691c0f9c5099b01f3')
df.save()
# create replica
base_url = 'file://' + settings.GET_FULL_STAGING_PATH_TEST
location = Location.load_location({
'name': 'staging-test-yyy', 'url': base_url, 'type': 'external',
'priority': 10, 'transfer_provider': 'local'})
replica = models.Replica(datafile=df, url='file://'+self.file,
protocol="staging",location=location)
replica.verify()
replica.save()
self.replica = replica
def setUp(self):
self.user = generate_user('fred')
Location.force_initialize()
self.experiment = generate_experiment(users=[self.user])
self.dataset = generate_dataset(experiments=[self.experiment])
self.server = SimpleHttpTestServer()
self.server.start()
def testMigrateStoreWithSpaces(self):
dest = Location.get_location('test')
local = Location.get_location('local')
datafile, replica = generate_datafile('1/1/Hi Mum', self.dataset,
"Hi mum")
datafile2, replica2 = generate_datafile('1/1/Hi Dad', self.dataset,
"Hi dad")
path = datafile.get_absolute_filepath()
self.assertTrue(os.path.exists(path))
path2 = datafile.get_absolute_filepath()
self.assertTrue(os.path.exists(path2))
# Migrate them
migrate_replica(replica, dest)
self.assertFalse(os.path.exists(path))
migrate_replica(replica2, dest)
self.assertFalse(os.path.exists(path2))
# Bring them back
migrate_replica(datafile.get_preferred_replica(), local)
self.assertTrue(os.path.exists(path))
migrate_replica(datafile2.get_preferred_replica(), local)
self.assertTrue(os.path.exists(path2))
def testReplicaVerify(self):
from django.conf import settings
saved = settings.REQUIRE_DATAFILE_CHECKSUMS
try:
Location.get_location('test')
datafile, replica = generate_datafile("1/2/3", self.dataset,
"Hi mum")
settings.REQUIRE_DATAFILE_CHECKSUMS = True
self.assertTrue(replica.verify(), 'Replica.verify() failed.')
replica.datafile.sha512sum = ''
replica.datafile.md5sum = ''
self.assertFalse(
replica.verify(),
'Replica.verify() succeeded despite no checksum '
'(settings.REQUIRE_DATAFILE_CHECKSUMS=True).')
self.assertFalse(replica.verify(allowEmptyChecksums=False),
'Replica.verify() succeeded despite no checksum '
'(allowEmptyChecksums=False)')
settings.REQUIRE_DATAFILE_CHECKSUMS = False
datafile.sha512sum = None
datafile.md5sum = None
self.assertTrue(replica.verify(allowEmptyChecksums=True),
'Replica.verify() failed wrongly '
'(allowEmptyChecksums=True)')
datafile.sha512sum = None
datafile.md5sum = None
self.assertTrue(replica.verify(),
'Replica.verify() failed wrongly')
finally:
settings.REQUIRE_DATAFILE_CHECKSUMS = saved
def setUp(self):
# Create test owner without enough details
username, email, password = ('testuser',
'*****@*****.**',
'password')
user = User.objects.create_user(username, email, password)
profile = UserProfile(user=user, isDjangoAccount=True)
profile.save()
Location.force_initialize()
# Create test experiment and make user the owner of it
experiment = Experiment(title='Text Experiment',
institution_name='Test Uni',
created_by=user)
experiment.save()
acl = ObjectACL(
pluginId='django_user',
entityId=str(user.id),
content_object=experiment,
canRead=True,
canWrite=True,
canDelete=True,
isOwner=True,
aclOwnershipType=ObjectACL.OWNER_OWNED)
acl.save()
dataset = Dataset(description='dataset description...')
dataset.save()
dataset.experiments.add(experiment)
dataset.save()
def create_datafile(filename):
testfile = path.join(path.dirname(__file__), 'fixtures',
filename)
size, sha512sum = get_size_and_sha512sum(testfile)
datafile = Dataset_File(dataset=dataset,
filename=path.basename(testfile),
size=size,
sha512sum=sha512sum)
datafile.save()
base_url = 'file://' + path.abspath(path.dirname(testfile))
location = Location.load_location({
'name': 'test-grabber', 'url': base_url, 'type': 'external',
'priority': 10, 'transfer_provider': 'local'})
replica = Replica(datafile=datafile,
url='file://'+path.abspath(testfile),
protocol='file',
location=location)
replica.verify()
replica.save()
return Dataset_File.objects.get(pk=datafile.pk)
self.dataset = dataset
self.datafiles = [create_datafile('data_grabber_test1.admin'),
create_datafile('testfile.txt')
]
def setUp(self):
from os import path, mkdir
from tempfile import mkdtemp
user = '******'
pwd = 'secret'
email = ''
self.user = User.objects.create_user(user, email, pwd)
self.userProfile = UserProfile(user=self.user).save()
self.test_dir = mkdtemp()
Location.force_initialize()
self.exp = Experiment(title='test exp1',
institution_name='monash',
created_by=self.user)
self.exp.save()
acl = ObjectACL(
pluginId=django_user,
entityId=str(self.user.id),
content_object=self.exp,
canRead=True,
isOwner=True,
aclOwnershipType=ObjectACL.OWNER_OWNED,
)
acl.save()
self.dataset = \
Dataset(description='dataset description...')
self.dataset.save()
self.dataset.experiments.add(self.exp)
self.dataset.save()
self.experiment_path = path.join(
settings.FILE_STORE_PATH,
str(self.dataset.get_first_experiment().id))
self.dataset_path = path.join(self.experiment_path,
str(self.dataset.id))
if not path.exists(self.experiment_path):
mkdir(self.experiment_path)
if not path.exists(self.dataset_path):
mkdir(self.dataset_path)
# write test file
self.filename = 'testfile.txt'
self.f1 = open(path.join(self.test_dir, self.filename), 'w')
self.f1.write('Test file 1')
self.f1.close()
self.f1_size = path.getsize(path.join(self.test_dir, self.filename))
self.f1 = open(path.join(self.test_dir, self.filename), 'r')
def setUp(self):
# Create test owner without enough details
username, email, password = ('testuser', '*****@*****.**',
'password')
user = User.objects.create_user(username, email, password)
profile = UserProfile(user=user, isDjangoAccount=True)
profile.save()
Location.force_initialize()
# Create test experiment and make user the owner of it
experiment = Experiment(title='Text Experiment',
institution_name='Test Uni',
created_by=user)
experiment.save()
acl = ObjectACL(
pluginId='django_user',
entityId=str(user.id),
content_object=experiment,
canRead=True,
isOwner=True,
aclOwnershipType=ObjectACL.OWNER_OWNED,
)
acl.save()
dataset = Dataset(description='dataset description...')
dataset.save()
dataset.experiments.add(experiment)
dataset.save()
def create_datafile(index):
testfile = path.join(path.dirname(__file__), 'fixtures',
'jeol_sem_test%d.txt' % index)
size, sha512sum = get_size_and_sha512sum(testfile)
datafile = Dataset_File(dataset=dataset,
filename=path.basename(testfile),
size=size,
sha512sum=sha512sum)
datafile.save()
base_url = 'file://' + path.abspath(path.dirname(testfile))
location = Location.load_location({
'name': 'test-jeol',
'url': base_url,
'type': 'external',
'priority': 10,
'transfer_provider': 'local'
})
replica = Replica(datafile=datafile,
url='file://' + path.abspath(testfile),
protocol='file',
location=location)
replica.verify()
replica.save()
return Dataset_File.objects.get(pk=datafile.pk)
self.dataset = dataset
self.datafiles = [create_datafile(i) for i in (1, 2)]
def setUp(self):
from django.contrib.auth.models import User
from tardis.tardis_portal.models import Location
user = '******'
pwd = 'secret'
email = ''
self.user = User.objects.create_user(user, email, pwd)
Location.force_initialize()
def setUp(self):
self.user = generate_user('fred')
Location.force_initialize()
self.experiment = generate_experiment(
users=[self.user],
title='Meanwhile, down in the archives ...',
url='http://example.com/something')
self.dataset = generate_dataset(experiments=[self.experiment])
def setUp(self):
from django.contrib.auth.models import User
from tardis.tardis_portal.models import Location
user = '******'
pwd = 'secret'
email = ''
self.user = User.objects.create_user(user, email, pwd)
Location.force_initialize()
def setUp(self):
from os import path, mkdir
from tempfile import mkdtemp
user = '******'
pwd = 'secret'
email = ''
self.user = User.objects.create_user(user, email, pwd)
self.userProfile = UserProfile(user=self.user).save()
self.test_dir = mkdtemp()
Location.force_initialize()
self.exp = Experiment(title='test exp1',
institution_name='monash', created_by=self.user)
self.exp.save()
acl = ObjectACL(
pluginId=django_user,
entityId=str(self.user.id),
content_object=self.exp,
canRead=True,
isOwner=True,
aclOwnershipType=ObjectACL.OWNER_OWNED,
)
acl.save()
self.dataset = \
Dataset(description='dataset description...')
self.dataset.save()
self.dataset.experiments.add(self.exp)
self.dataset.save()
self.experiment_path = path.join(settings.FILE_STORE_PATH,
str(self.dataset.get_first_experiment().id))
self.dataset_path = path.join(self.experiment_path,
str(self.dataset.id))
if not path.exists(self.experiment_path):
mkdir(self.experiment_path)
if not path.exists(self.dataset_path):
mkdir(self.dataset_path)
# write test file
self.filename = 'testfile.txt'
self.f1 = open(path.join(self.test_dir, self.filename), 'w')
self.f1.write('Test file 1')
self.f1.close()
self.f1_size = path.getsize(path.join(self.test_dir,
self.filename))
self.f1 = open(path.join(self.test_dir, self.filename), 'r')
def _infer_location(path):
if urlparse.urlparse(path).scheme == '':
loc = Location.get_default_location()
else:
loc = Location.get_location_for_url(path)
if loc:
return loc
else:
raise Exception('Cannot infer a location for %s' % path)
def _infer_location(path):
if urlparse.urlparse(path).scheme == '':
loc = Location.get_default_location()
else:
loc = Location.get_location_for_url(path)
if loc:
return loc
else:
raise Exception('Cannot infer a location for %s' % path)
def testRemoveExperimentData(self):
# First with no sharing
self._build()
archive_location = Location.get_location('archtest')
try:
nos_experiments = Experiment.objects.count()
nos_datasets = Dataset.objects.count()
nos_datafiles = Dataset_File.objects.count()
nos_replicas = Replica.objects.count()
self.assertTrue(exists(self.replica.get_absolute_filepath()))
remove_experiment_data(self.experiment,
'http://example.com/some.tar.gz',
archive_location)
self.assertEquals(nos_experiments, Experiment.objects.count())
self.assertEquals(nos_datasets, Dataset.objects.count())
self.assertEquals(nos_datafiles, Dataset_File.objects.count())
self.assertEquals(nos_replicas, Replica.objects.count())
new_replica = self.datafile.get_preferred_replica()
self.assertTrue(self.replica.id != new_replica.id)
self.assertFalse(new_replica.stay_remote)
self.assertTrue(new_replica.verified)
self.assertEqual(self.replica.protocol, new_replica.protocol)
self.assertEqual(archive_location.id, new_replica.location.id)
self.assertEqual('http://example.com/some.tar.gz#1/1/1',
new_replica.url)
self.assertFalse(exists(self.replica.get_absolute_filepath()))
finally:
self._clear()
# (Check that the deletes cascaded ... )
self.assertEquals(0, Dataset_File.objects.count())
self.assertEquals(0, Replica.objects.count())
# Repeat, but with the first dataset in 2 experiments.
self._build()
self.dataset.experiments.add(self.experiment2)
archive_location = Location.get_location('archtest')
try:
nos_experiments = Experiment.objects.count()
nos_datasets = Dataset.objects.count()
nos_datafiles = Dataset_File.objects.count()
nos_replicas = Replica.objects.count()
self.assertTrue(exists(self.replica.get_absolute_filepath()))
remove_experiment_data(self.experiment,
'http://example.com/some.tar.gz',
archive_location)
self.assertEquals(nos_experiments, Experiment.objects.count())
self.assertEquals(nos_datasets, Dataset.objects.count())
self.assertEquals(nos_datafiles, Dataset_File.objects.count())
self.assertEquals(nos_replicas, Replica.objects.count())
new_replica = self.datafile.get_preferred_replica()
self.assertTrue(self.replica.id == new_replica.id)
self.assertTrue(exists(self.replica.get_absolute_filepath()))
self.assertFalse(exists(self.replica2.get_absolute_filepath()))
finally:
self._clear()
def setUpClass(cls):
cls.priorcwd = os.getcwd()
os.chdir(os.path.dirname(__file__)+'/atom_test')
cls.server = TestWebServer()
cls.server.start()
Location.force_initialize()
Location.load_location({
'name': 'test-atom',
'transfer_provider': 'http',
'url': 'http://localhost:4272/files/',
'type': 'external',
'priority': 10})
Location.load_location({
'name': 'test-atom2',
'transfer_provider': 'http',
'url': 'http://mydatagrabber.cmm.uq.edu.au/files',
'type': 'external',
'priority': 10})
files = path.realpath(path.join(path.dirname(__file__),
'atom_test', 'files'))
Location.load_location({
'name': 'test-atom3',
'transfer_provider': 'local',
'url': 'file://' + files,
'type': 'external',
'priority': 10})
def setUpClass(cls):
cls.priorcwd = os.getcwd()
os.chdir(os.path.dirname(__file__) + '/atom_test')
cls.server = TestWebServer()
cls.server.start()
Location.force_initialize()
Location.load_location({
'name': 'test-atom',
'transfer_provider': 'http',
'url': 'http://localhost:4272/files/',
'type': 'external',
'priority': 10
})
Location.load_location({
'name': 'test-atom2',
'transfer_provider': 'http',
'url': 'http://mydatagrabber.cmm.uq.edu.au/files',
'type': 'external',
'priority': 10
})
files = path.realpath(
path.join(path.dirname(__file__), 'atom_test', 'files'))
Location.load_location({
'name': 'test-atom3',
'transfer_provider': 'local',
'url': 'file://' + files,
'type': 'external',
'priority': 10
})
def testScoring(self):
self._setup()
scorer = MigrationScorer(Location.get_location('local').id)
self.assertEquals(2.0, scorer.datafile_score(self.df1))
self.assertEquals(2, get_user_priority(self.user1))
self.assertEquals(1, get_user_priority(self.user2))
self.assertEquals(1.0, scorer.user_score(self.user1))
self.assertEquals(2.0, scorer.user_score(self.user2))
self.assertEquals(2.0, scorer.experiment_score(self.exp1))
self.assertEquals(2.0, scorer.dataset_score(self.df1.dataset))
self.assertEquals(4.0, scorer.score_datafile(self.df1))
self.assertEquals([(self.df1, self.rep1, 4.0)],
scorer.score_datafiles_in_dataset(self.ds1))
self.assertEquals([(self.df5, self.rep5, 8.0),
(self.df4, self.rep4, 6.0),
(self.df1, self.rep1, 4.0)],
scorer.score_datafiles_in_experiment(self.exp1))
self.assertEquals([(self.df5, self.rep5, 8.0),
(self.df4, self.rep4, 6.0)],
scorer.score_datafiles_in_experiment(self.exp2))
self.assertEquals([(self.df6, self.rep6, 5.0)],
scorer.score_datafiles_in_experiment(self.exp3))
self.assertEquals([(self.df5, self.rep5, 8.0),
(self.df4, self.rep4, 6.0),
(self.df6, self.rep6, 5.0),
(self.df1, self.rep1, 4.0),
(self.df7, self.rep7, 0.0),
(self.df8, self.rep8, 0.0)],
scorer.score_all_datafiles())
self.assertEquals([(self.df7, self.rep7, 0.0),
(self.df8, self.rep8, 0.0)],
scorer.score_datafiles_in_dataset(self.ds4))
def testScoringWithTimes(self):
self._setup()
scorer = MigrationScorer(
Location.get_location('local').id, {
'user_priority_weighting': [5.0, 2.0, 1.0, 0.5, 0.2],
'file_size_weighting': 1.0,
'file_access_weighting': 1.0,
'file_age_weighting': 1.0,
'file_size_threshold': 0,
'file_access_threshold': 0,
'file_age_threshold': 1})
self.assertEquals(0.0, scorer.datafile_score(self.df1))
f = tempfile.NamedTemporaryFile(dir=settings.FILE_STORE_PATH)
f.write("Hi Mom!!\n")
rep = Replica.objects.get(pk=self.rep1.pk)
rep.url = f.name
rep.save()
self.assertEquals(2.0, scorer.datafile_score(self.df1))
older = time.time() - (60 * 60 * 24 + 300)
os.utime(f.name, (older, older))
self.assertEquals(3.0, scorer.datafile_score(self.df1))
older = time.time() - (60 * 60 * 24 * 2 + 300)
os.utime(f.name, (older, older))
self.assertEquals(5.0, scorer.datafile_score(self.df1))
f.close()
def stage_replica(replica):
from django.core.files.uploadedfile import TemporaryUploadedFile
from tardis.tardis_portal.models import Replica, Location
if not replica.location.type == 'external':
raise ValueError('Only external replicas can be staged')
if getattr(settings, "DEEP_DATASET_STORAGE", False):
relurl = path.relpath(replica.url[7:], settings.SYNC_TEMP_PATH)
spliturl = relurl.split(os.sep)[1:]
subdir = path.dirname(path.join(*spliturl))
else:
subdir = None
with TemporaryUploadedFile(replica.datafile.filename, None, None,
None) as tf:
if replica.verify(tempfile=tf.file):
if not replica.stay_remote:
tf.file.flush()
target_replica = {
'datafile':
replica.datafile,
'url':
write_uploaded_file_to_dataset(replica.datafile.dataset,
tf,
subdir=subdir),
'location':
Location.get_default_location(),
'verified':
True,
'protocol':
''
}
Replica.objects.filter(id=replica.id).update(**target_replica)
return True
else:
return False
def remove_experiment_data(exp, archive_url, archive_location):
"""Remove the online Replicas for an Experiment that are not shared with
other Experiments. When Replicas are removed, they are replaced with
offline replicas whose 'url' consists of the archive_url, with the
archive pathname for the datafile as a url fragment id.
"""
for ds in Dataset.objects.filter(experiments=exp):
if ds.experiments.count() == 1:
for df in Dataset_File.objects.filter(dataset=ds):
replicas = Replica.objects.filter(datafile=df,
location__type='online')
if replicas.count() > 0:
for replica in replicas:
location = Location.get_location(replica.location.name)
location.provider.remove_file(replica)
if archive_url:
old_replica = replicas[0]
path_in_archive = '%s/%s/%s' % (
exp.id, ds.id, df.filename)
new_replica_url = '%s#%s' % (
archive_url, quote(path_in_archive))
new_replica = Replica(datafile=old_replica.datafile,
url=new_replica_url,
protocol=old_replica.protocol,
verified=True,
stay_remote=False,
location=archive_location)
new_replica.save()
replicas.delete()
def testScoring(self):
self._setup()
scorer = MigrationScorer(Location.get_location('local').id)
self.assertEquals(2.0, scorer.datafile_score(self.df1))
self.assertEquals(2, get_user_priority(self.user1))
self.assertEquals(1, get_user_priority(self.user2))
self.assertEquals(1.0, scorer.user_score(self.user1))
self.assertEquals(2.0, scorer.user_score(self.user2))
self.assertEquals(2.0, scorer.experiment_score(self.exp1))
self.assertEquals(2.0, scorer.dataset_score(self.df1.dataset))
self.assertEquals(4.0, scorer.score_datafile(self.df1))
self.assertEquals([(self.df1, self.rep1, 4.0)],
scorer.score_datafiles_in_dataset(self.ds1))
self.assertEquals([(self.df5, self.rep5, 8.0),
(self.df4, self.rep4, 6.0),
(self.df1, self.rep1, 4.0)],
scorer.score_datafiles_in_experiment(self.exp1))
self.assertEquals([(self.df5, self.rep5, 8.0),
(self.df4, self.rep4, 6.0)],
scorer.score_datafiles_in_experiment(self.exp2))
self.assertEquals([(self.df6, self.rep6, 5.0)],
scorer.score_datafiles_in_experiment(self.exp3))
self.assertEquals([(self.df5, self.rep5, 8.0),
(self.df4, self.rep4, 6.0),
(self.df6, self.rep6, 5.0),
(self.df1, self.rep1, 4.0),
(self.df7, self.rep7, 0.0),
(self.df8, self.rep8, 0.0)],
scorer.score_all_datafiles())
self.assertEquals([(self.df7, self.rep7, 0.0),
(self.df8, self.rep8, 0.0)],
scorer.score_datafiles_in_dataset(self.ds4))
def testScoringWithTimes(self):
self._setup()
scorer = MigrationScorer(
Location.get_location('local').id, {
'user_priority_weighting': [5.0, 2.0, 1.0, 0.5, 0.2],
'file_size_weighting': 1.0,
'file_access_weighting': 1.0,
'file_age_weighting': 1.0,
'file_size_threshold': 0,
'file_access_threshold': 0,
'file_age_threshold': 1
})
self.assertEquals(0.0, scorer.datafile_score(self.df1))
f = tempfile.NamedTemporaryFile(dir=settings.FILE_STORE_PATH)
f.write("Hi Mom!!\n")
rep = Replica.objects.get(pk=self.rep1.pk)
rep.url = f.name
rep.save()
self.assertEquals(2.0, scorer.datafile_score(self.df1))
older = time.time() - (60 * 60 * 24 + 300)
os.utime(f.name, (older, older))
self.assertEquals(3.0, scorer.datafile_score(self.df1))
older = time.time() - (60 * 60 * 24 * 2 + 300)
os.utime(f.name, (older, older))
self.assertEquals(5.0, scorer.datafile_score(self.df1))
f.close()
def testLocalFile(self):
content = urandom(1024)
cf = ContentFile(content, 'background_task_testfile')
# Create new Datafile
datafile = Dataset_File(dataset=self.dataset)
datafile.filename = cf.name
datafile.size = len(content)
datafile.sha512sum = hashlib.sha512(content).hexdigest()
datafile.save()
replica = Replica(datafile=datafile,
url=write_uploaded_file_to_dataset(self.dataset, cf),
location=Location.get_default_location())
replica.save()
def get_replica(datafile):
return Replica.objects.get(datafile=datafile)
# undo auto-verify:
replica.verified = False
replica.save(update_fields=['verified'])
# Check that it's not currently verified
expect(get_replica(datafile).verified).to_be(False)
# Check it verifies
verify_files()
expect(get_replica(datafile).verified).to_be(True)
def stage_replica(replica):
from django.core.files.uploadedfile import TemporaryUploadedFile
from tardis.tardis_portal.models import Replica, Location
if not replica.location.type == 'external':
raise ValueError('Only external replicas can be staged')
if getattr(settings, "DEEP_DATASET_STORAGE", False):
relurl = path.relpath(replica.url[7:], settings.SYNC_TEMP_PATH)
spliturl = relurl.split(os.sep)[1:]
subdir = path.dirname(path.join(*spliturl))
else:
subdir = None
with TemporaryUploadedFile(replica.datafile.filename,
None, None, None) as tf:
if replica.verify(tempfile=tf.file):
if not replica.stay_remote:
tf.file.flush()
target_replica = {
'datafile': replica.datafile,
'url': write_uploaded_file_to_dataset(
replica.datafile.dataset, tf,
subdir=subdir),
'location': Location.get_default_location(),
'verified': True,
'protocol': ''}
Replica.objects.filter(id=replica.id).update(**target_replica)
return True
else:
return False
def create_datafile(index):
testfile = path.join(path.dirname(__file__), 'fixtures',
'jeol_sem_test%d.txt' % index)
size, sha512sum = get_size_and_sha512sum(testfile)
datafile = Dataset_File(dataset=dataset,
filename=path.basename(testfile),
size=size,
sha512sum=sha512sum)
datafile.save()
base_url = 'file://' + path.abspath(path.dirname(testfile))
location = Location.load_location({
'name': 'test-jeol',
'url': base_url,
'type': 'external',
'priority': 10,
'transfer_provider': 'local'
})
replica = Replica(datafile=datafile,
url='file://' + path.abspath(testfile),
protocol='file',
location=location)
replica.verify()
replica.save()
return Dataset_File.objects.get(pk=datafile.pk)
def process_enclosure(self, dataset, enclosure):
filename = getattr(enclosure, 'title', basename(enclosure.href))
datafile = Dataset_File(filename=filename, dataset=dataset)
try:
datafile.mimetype = enclosure.mime
except AttributeError:
pass
try:
datafile.size = enclosure.length
except AttributeError:
pass
try:
hash = enclosure.hash
# Split on white space, then ':' to get tuples to feed into dict
hashdict = dict([s.partition(':')[::2] for s in hash.split()])
# Set SHA-512 sum
datafile.sha512sum = hashdict['sha-512']
except AttributeError:
pass
datafile.save()
url = enclosure.href
# This means we will allow the atom feed to feed us any enclosure
# URL that matches a registered location. Maybe we should restrict
# this to a specific location.
location = Location.get_location_for_url(url)
if not location:
logger.error('Rejected ingestion for unknown location %s' % url)
return
replica = Replica(datafile=datafile, url=url,
location=location)
replica.protocol = enclosure.href.partition('://')[0]
replica.save()
self.make_local_copy(replica)
def process_enclosure(self, dataset, enclosure):
filename = getattr(enclosure, 'title', basename(enclosure.href))
datafile = Dataset_File(filename=filename, dataset=dataset)
try:
datafile.mimetype = enclosure.mime
except AttributeError:
pass
try:
datafile.size = enclosure.length
except AttributeError:
pass
try:
hash = enclosure.hash
# Split on white space, then ':' to get tuples to feed into dict
hashdict = dict([s.partition(':')[::2] for s in hash.split()])
# Set SHA-512 sum
datafile.sha512sum = hashdict['sha-512']
except AttributeError:
pass
datafile.save()
url = enclosure.href
# This means we will allow the atom feed to feed us any enclosure
# URL that matches a registered location. Maybe we should restrict
# this to a specific location.
location = Location.get_location_for_url(url)
if not location:
logger.error('Rejected ingestion for unknown location %s' % url)
return
replica = Replica(datafile=datafile, url=url, location=location)
replica.protocol = enclosure.href.partition('://')[0]
replica.save()
self.make_local_copy(replica)
def testMigrateRestore(self):
dest = Location.get_location('test')
local = Location.get_location('local')
datafile, replica = generate_datafile(None,
self.dataset,
"Hi mum",
verify=False,
verify_checksums_req=True)
# Attempt to migrate without datafile hashes ... should
# fail because we can't verify.
with self.assertRaises(MigrationError):
migrate_replica(replica, dest)
# Verify sets hashes ...
self.assertEquals(replica.verify(allowEmptyChecksums=True), True)
replica = Replica.objects.get(pk=replica.pk)
path = datafile.get_absolute_filepath()
self.assertTrue(os.path.exists(path))
self.assertTrue(migrate_replica(replica, dest))
self.assertFalse(os.path.exists(path))
# Bring it back
new_replica = datafile.get_preferred_replica()
url = new_replica.url
self.assertTrue(migrate_replica(new_replica, local))
self.assertTrue(os.path.exists(path))
# Check it was deleted remotely
with self.assertRaises(TransferError):
dest.provider.get_length(new_replica)
# Refresh the datafile object because it is now stale ...
datafile = Dataset_File.objects.get(id=datafile.id)
replica = datafile.get_preferred_replica()
# Repeat the process with 'noRemove'
self.assertTrue(migrate_replica(replica, dest, noRemove=True))
new_replica = datafile.get_preferred_replica()
self.assertTrue(os.path.exists(path))
self.assertEquals(dest.provider.get_length(new_replica), 6)
migrate_replica(new_replica, local, noRemove=True)
newpath = datafile.get_absolute_filepath()
replica = datafile.get_preferred_replica()
self.assertTrue(os.path.exists(path))
self.assertTrue(os.path.exists(newpath))
self.assertNotEqual(path, newpath)
self.assertEquals(dest.provider.get_length(new_replica), 6)
def testMigrateRestore(self):
dest = Location.get_location('test')
local = Location.get_location('local')
datafile, replica = generate_datafile(None, self.dataset, "Hi mum",
verify=False,
verify_checksums_req=True)
# Attempt to migrate without datafile hashes ... should
# fail because we can't verify.
with self.assertRaises(MigrationError):
migrate_replica(replica, dest)
# Verify sets hashes ...
self.assertEquals(replica.verify(allowEmptyChecksums=True), True)
replica = Replica.objects.get(pk=replica.pk)
path = datafile.get_absolute_filepath()
self.assertTrue(os.path.exists(path))
self.assertTrue(migrate_replica(replica, dest))
self.assertFalse(os.path.exists(path))
# Bring it back
new_replica = datafile.get_preferred_replica()
url = new_replica.url
self.assertTrue(migrate_replica(new_replica, local))
self.assertTrue(os.path.exists(path))
# Check it was deleted remotely
with self.assertRaises(TransferError):
dest.provider.get_length(new_replica)
# Refresh the datafile object because it is now stale ...
datafile = Dataset_File.objects.get(id=datafile.id)
replica = datafile.get_preferred_replica()
# Repeat the process with 'noRemove'
self.assertTrue(migrate_replica(replica, dest, noRemove=True))
new_replica = datafile.get_preferred_replica()
self.assertTrue(os.path.exists(path))
self.assertEquals(dest.provider.get_length(new_replica), 6)
migrate_replica(new_replica, local, noRemove=True)
newpath = datafile.get_absolute_filepath()
replica = datafile.get_preferred_replica()
self.assertTrue(os.path.exists(path))
self.assertTrue(os.path.exists(newpath))
self.assertNotEqual(path, newpath)
self.assertEquals(dest.provider.get_length(new_replica), 6)
def setUp(self):
# Create test owner without enough details
username, email, password = ('testuser',
'*****@*****.**',
'password')
user = User.objects.create_user(username, email, password)
profile = UserProfile(user=user, isDjangoAccount=True)
profile.save()
# Need UserAuthentication
UserAuthentication(userProfile=profile,
username=username,
authenticationMethod='localdb').save()
# Create staging dir
from os import path, makedirs
staging_dir = path.join(settings.STAGING_PATH, username)
if not path.exists(staging_dir):
makedirs(staging_dir)
# Ensure that staging dir is set up properly
expect(get_full_staging_path(username)).to_be_truthy()
Location.force_initialize()
# Create test experiment and make user the owner of it
experiment = Experiment(title='Text Experiment',
institution_name='Test Uni',
created_by=user)
experiment.save()
acl = ExperimentACL(
pluginId=django_user,
entityId=str(user.id),
experiment=experiment,\
canRead=True,
isOwner=True,
aclOwnershipType=ExperimentACL.OWNER_OWNED,
)
acl.save()
self.dataset = \
Dataset(description='dataset description...')
self.dataset.save()
self.dataset.experiments.add(experiment)
self.dataset.save()
self.username, self.password = (username, password)
def do_ext_provider(self, loc_name):
# This test requires an external test server configured
# as per the 'dest_name' destination. We skip the test is the
# server doesn't respond.
loc = Location.get_location(loc_name)
if loc.provider.alive():
self.do_provider(loc)
else:
print "SKIPPING TEST - %s server on %s is not responding\n" % (loc_name, loc.url)
def testArchiveExperiment(self):
dataset = generate_dataset()
experiment = generate_experiment([dataset], [self.dummy_user])
datafile, _ = generate_datafile(None, dataset, "Hi grandpa")
archtest = Location.get_location('archtest')
# Dry run ...
out = StringIO()
try:
call_command('archive', experiment.id,
verbosity=1, stdout=out, dryRun=True)
except SystemExit:
pass
out.seek(0)
self.assertEquals(out.read(),
'Would have archived experiment %s\n' % experiment.id)
# Dry run ... all
out = StringIO()
try:
call_command('archive', all=True,
verbosity=1, stdout=out, dryRun=True)
except SystemExit:
pass
out.seek(0)
self.assertEquals(out.read(),
'Would have archived experiment %s\n' % experiment.id)
# Do one ... to file
out = StringIO()
try:
call_command('archive', experiment.id, directory='/tmp',
verbosity=1, stdout=out)
except SystemExit:
pass
out.seek(0)
self.assertEquals(
out.read(),
'Archived experiment %s to /tmp/%s-archive.tar.gz\n' \
'Archived 1 experiments with 0 errors\n' % \
(experiment.id, experiment.id))
# Do one ... to archtest
out = StringIO()
try:
call_command('archive', experiment.id, location='archtest',
verbosity=1, stdout=out)
except SystemExit:
pass
out.seek(0)
self.assertEquals(
out.read(),
'Archived experiment %s to %s%s-archive.tar.gz\n' \
'Archived 1 experiments with 0 errors\n' % \
(experiment.id, archtest.provider.base_url,
experiment.id))
def do_ext_provider(self, loc_name):
# This test requires an external test server configured
# as per the 'dest_name' destination. We skip the test is the
# server doesn't respond.
loc = Location.get_location(loc_name)
if loc.provider.alive():
self.do_provider(loc)
else:
print 'SKIPPING TEST - %s server on %s is not responding\n' % \
(loc_name, loc.url)
def setUp(self):
# Create test owner without enough details
username, email, password = ('testuser', '*****@*****.**',
'password')
user = User.objects.create_user(username, email, password)
profile = UserProfile(user=user, isDjangoAccount=True)
profile.save()
# Need UserAuthentication
UserAuthentication(userProfile=profile,
username=username,
authenticationMethod='localdb').save()
# Create staging dir
from os import path, makedirs
staging_dir = path.join(settings.STAGING_PATH, username)
if not path.exists(staging_dir):
makedirs(staging_dir)
# Ensure that staging dir is set up properly
expect(get_full_staging_path(username)).to_be_truthy()
Location.force_initialize()
# Create test experiment and make user the owner of it
experiment = Experiment(title='Text Experiment',
institution_name='Test Uni',
created_by=user)
experiment.save()
acl = ObjectACL(
pluginId=django_user,
entityId=str(user.id),
content_object=experiment,
canRead=True,
isOwner=True,
aclOwnershipType=ObjectACL.OWNER_OWNED,
)
acl.save()
self.dataset = \
Dataset(description='dataset description...')
self.dataset.save()
self.dataset.experiments.add(experiment)
self.dataset.save()
self.username, self.password = (username, password)
def _get_destination(self, destName, default):
if not destName:
if not default:
raise CommandError("No default destination configured")
else:
destName = default
dest = Location.get_location(destName)
if not dest:
raise CommandError("Destination %s not known" % destName)
return dest
def _build(dataset, filename, url, protocol):
from tardis.tardis_portal.models import \
Dataset_File, Replica, Location
datafile = Dataset_File(dataset=dataset, filename=filename)
datafile.save()
replica = Replica(datafile=datafile, url=url,
protocol=protocol,
location=Location.get_default_location())
replica.save()
return datafile
def _build(dataset, filename, url, protocol):
from tardis.tardis_portal.models import \
Dataset_File, Replica, Location
datafile = Dataset_File(dataset=dataset, filename=filename)
datafile.save()
replica = Replica(datafile=datafile, url=url,
protocol=protocol,
location=Location.get_default_location())
replica.save()
return datafile
def _create_datafile():
user = User.objects.create_user('testuser', '*****@*****.**', 'pwd')
user.save()
UserProfile(user=user).save()
Location.force_initialize()
full_access = Experiment.PUBLIC_ACCESS_FULL
experiment = Experiment.objects.create(title="IIIF Test",
created_by=user,
public_access=full_access)
experiment.save()
ObjectACL(content_object=experiment,
pluginId='django_user',
entityId=str(user.id),
isOwner=True,
canRead=True,
canWrite=True,
canDelete=True,
aclOwnershipType=ObjectACL.OWNER_OWNED).save()
dataset = Dataset()
dataset.save()
dataset.experiments.add(experiment)
dataset.save()
# Create new Datafile
tempfile = TemporaryUploadedFile('iiif_stored_file', None, None, None)
with Image(filename='magick:rose') as img:
img.format = 'tiff'
img.save(file=tempfile.file)
tempfile.file.flush()
datafile = Dataset_File(dataset=dataset,
size=os.path.getsize(tempfile.file.name),
filename='iiif_named_file')
replica = Replica(datafile=datafile,
url=write_uploaded_file_to_dataset(dataset, tempfile),
location=Location.get_default_location())
replica.verify(allowEmptyChecksums=True)
datafile.save()
replica.datafile = datafile
replica.save()
return datafile
def _create_datafile():
user = User.objects.create_user("testuser", "*****@*****.**", "pwd")
user.save()
UserProfile(user=user).save()
Location.force_initialize()
full_access = Experiment.PUBLIC_ACCESS_FULL
experiment = Experiment.objects.create(title="IIIF Test", created_by=user, public_access=full_access)
experiment.save()
ObjectACL(
content_object=experiment,
pluginId="django_user",
entityId=str(user.id),
isOwner=True,
canRead=True,
canWrite=True,
canDelete=True,
aclOwnershipType=ObjectACL.OWNER_OWNED,
).save()
dataset = Dataset()
dataset.save()
dataset.experiments.add(experiment)
dataset.save()
# Create new Datafile
tempfile = TemporaryUploadedFile("iiif_stored_file", None, None, None)
with Image(filename="magick:rose") as img:
img.format = "tiff"
img.save(file=tempfile.file)
tempfile.file.flush()
datafile = Dataset_File(dataset=dataset, size=os.path.getsize(tempfile.file.name), filename="iiif_named_file")
replica = Replica(
datafile=datafile,
url=write_uploaded_file_to_dataset(dataset, tempfile),
location=Location.get_default_location(),
)
replica.verify(allowEmptyChecksums=True)
datafile.save()
replica.datafile = datafile
replica.save()
return datafile
def testMigrationNoHashes(self):
# Tweak the server to turn off the '?metadata' query
self.server.server.allowQuery = False
dest = Location.get_location('test')
datafile, replica = generate_datafile("1/2/3", self.dataset, "Hi mum")
self.assertEquals(replica.verify(allowEmptyChecksums=True), True)
path = datafile.get_absolute_filepath()
self.assertTrue(os.path.exists(path))
migrate_replica(replica, dest)
self.assertFalse(os.path.exists(path))
def testMigrationNoHashes(self):
# Tweak the server to turn off the '?metadata' query
self.server.server.allowQuery = False
dest = Location.get_location('test')
datafile, replica = generate_datafile("1/2/3", self.dataset, "Hi mum")
self.assertEquals(replica.verify(allowEmptyChecksums=True), True)
path = datafile.get_absolute_filepath()
self.assertTrue(os.path.exists(path))
migrate_replica(replica, dest)
self.assertFalse(os.path.exists(path))
def _build_datafile(self,
testfile,
filename,
dataset,
url,
protocol='',
checksum=None,
size=None,
mimetype=''):
filesize, sha512sum = get_size_and_sha512sum(testfile)
datafile = Dataset_File(
dataset=dataset,
filename=filename,
mimetype=mimetype,
size=str(size if size != None else filesize),
sha512sum=(checksum if checksum else sha512sum))
datafile.save()
if urlparse.urlparse(url).scheme == '':
location = Location.get_location('local')
else:
location = Location.get_location_for_url(url)
if not location:
location = Location.load_location({
'name':
filename,
'url':
urlparse.urljoin(url, '.'),
'type':
'external',
'priority':
10,
'transfer_provider':
'local'
})
replica = Replica(datafile=datafile,
protocol=protocol,
url=url,
location=location)
replica.verify()
replica.save()
return Dataset_File.objects.get(pk=datafile.pk)
def testProviderInstantiation(self):
"""
Test that transfer_provider instantiation works
"""
provider = Location.get_location("test").provider
self.assertIsInstance(provider, TransferProvider)
self.assertIsInstance(provider, SimpleHttpTransfer)
self.assertEqual(provider.base_url, "http://127.0.0.1:4272/data/")
provider = Location.get_location("test2").provider
self.assertIsInstance(provider, TransferProvider)
self.assertIsInstance(provider, WebDAVTransfer)
self.assertFalse(401 in provider.opener.handle_error["http"])
self.assertEqual(provider.base_url, "http://127.0.0.1/data2/")
provider = Location.get_location("test3").provider
self.assertIsInstance(provider, TransferProvider)
self.assertIsInstance(provider, WebDAVTransfer)
self.assertTrue(401 in provider.opener.handle_error["http"])
self.assertEqual(provider.base_url, "http://127.0.0.1/data3/")
def _create_test_dataset(nosDatafiles):
ds_ = Dataset(description='happy snaps of plumage')
ds_.save()
for i in range (0, nosDatafiles) :
df_ = Dataset_File(dataset=ds_, size='21', sha512sum='bogus')
df_.save()
rep_ = Replica(datafile=df_,
url='http://planet-python.org/' + str(_next_id()),
location=Location.get_default_location())
rep_.save()
ds_.save()
return ds_
def testProviderInstantiation(self):
'''
Test that transfer_provider instantiation works
'''
provider = Location.get_location('test').provider
self.assertIsInstance(provider, TransferProvider)
self.assertIsInstance(provider, SimpleHttpTransfer)
self.assertEqual(provider.base_url, 'http://127.0.0.1:4272/data/')
provider = Location.get_location('test2').provider
self.assertIsInstance(provider, TransferProvider)
self.assertIsInstance(provider, WebDAVTransfer)
self.assertFalse(401 in provider.opener.handle_error['http'])
self.assertEqual(provider.base_url, 'http://127.0.0.1/data2/')
provider = Location.get_location('test3').provider
self.assertIsInstance(provider, TransferProvider)
self.assertIsInstance(provider, WebDAVTransfer)
self.assertTrue(401 in provider.opener.handle_error['http'])
self.assertEqual(provider.base_url, 'http://127.0.0.1/data3/')
def _create_test_dataset(nosDatafiles):
ds_ = Dataset(description='happy snaps of plumage')
ds_.save()
for i in range(0, nosDatafiles):
df_ = Dataset_File(dataset=ds_, size='21', sha512sum='bogus')
df_.save()
rep_ = Replica(datafile=df_,
url='http://planet-python.org/' + str(_next_id()),
location=Location.get_default_location())
rep_.save()
ds_.save()
return ds_
def setUp(self):
"""
setting up essential objects, copied from tests above
"""
Location.force_initialize()
self.location = Location.get_location('local')
user = '******'
pwd = 'secret'
email = ''
self.user = User.objects.create_user(user, email, pwd)
self.userProfile = UserProfile(user=self.user).save()
self.exp = Experiment(title='test exp1',
institution_name='monash',
created_by=self.user)
self.exp.save()
self.acl = ObjectACL(
pluginId=django_user,
entityId=str(self.user.id),
content_object=self.exp,
canRead=True,
isOwner=True,
aclOwnershipType=ObjectACL.OWNER_OWNED,
)
self.acl.save()
self.dataset = Dataset(description='dataset description...')
self.dataset.save()
self.dataset.experiments.add(self.exp)
self.dataset.save()
self.dataset_file = Dataset_File(dataset=self.dataset,
size=42,
filename="foo",
md5sum="junk")
self.dataset_file.save()
self.replica = Replica(datafile=self.dataset_file,
url="http://foo",
location=self.location,
verified=False)
self.replica.save()
def _get_destination(self, destName, default):
if not destName:
if not default:
raise CommandError("No default destination configured")
else:
destName = default
try:
dest = Location.get_location(destName)
if not dest:
raise CommandError("Destination %s not known" % destName)
return dest
except MigrationError as e:
raise CommandError("Migration error: %s" % e.args[0])
def _get_destination(self, destName, default):
if not destName:
if not default:
raise CommandError("No default destination configured")
else:
destName = default
try:
dest = Location.get_location(destName)
if not dest:
raise CommandError("Destination %s not known" % destName)
return dest
except MigrationError as e:
raise CommandError("Migration error: %s" % e.args[0])
def _build_datafile(self, testfile, filename, dataset, url,
protocol='', checksum=None, size=None, mimetype=''):
filesize, sha512sum = get_size_and_sha512sum(testfile)
datafile = Dataset_File(dataset=dataset, filename=filename,
mimetype=mimetype,
size=str(size if size != None else filesize),
sha512sum=(checksum if checksum else sha512sum))
datafile.save()
if urlparse.urlparse(url).scheme == '':
location = Location.get_location('local')
else:
location = Location.get_location_for_url(url)
if not location:
location = Location.load_location({
'name': filename, 'url': urlparse.urljoin(url, '.'),
'type': 'external',
'priority': 10, 'transfer_provider': 'local'})
replica = Replica(datafile=datafile, protocol=protocol, url=url,
location=location)
replica.verify()
replica.save()
return Dataset_File.objects.get(pk=datafile.pk)
def _setup(self):
Location.force_initialize()
self.user1 = generate_user('joe', 2)
self.user2 = generate_user('fred', 1)
self.exp1 = generate_experiment(users=[self.user1, self.user2])
self.exp2 = generate_experiment(users=[self.user1])
self.exp3 = generate_experiment(users=[self.user1])
self.exp4 = generate_experiment(users=[self.user1])
self.ds1 = generate_dataset(experiments=[self.exp1])
self.ds2 = generate_dataset(experiments=[self.exp1, self.exp2])
self.ds3 = generate_dataset(experiments=[self.exp3])
self.ds4 = generate_dataset(experiments=[self.exp4])
self.df1, self.rep1 = generate_datafile('1/2/1', self.ds1, size=100)
self.df2, self.rep2 = generate_datafile('1/2/2',
self.ds1,
size=100,
verified=False)
self.df3, self.rep3 = generate_datafile(
'http://127.0.0.1:4272/data/1/2/3', self.ds1, size=1000)
self.df4, self.rep4 = generate_datafile('1/2/4', self.ds2, size=1000)
self.df5, self.rep5 = generate_datafile('1/2/5', self.ds2, size=10000)
self.df6, self.rep6 = generate_datafile('1/2/6', self.ds3, size=100000)
self.df7, self.rep7 = generate_datafile('1/2/7', self.ds4, size=0)
self.df8, self.rep8 = generate_datafile('1/2/8', self.ds4, size=-1)
def create_staging_datafile(filepath, username, dataset_id):
from tardis.tardis_portal.models import Dataset_File, Dataset, Replica, \
Location
dataset = Dataset.objects.get(id=dataset_id)
url, size = get_staging_url_and_size(username, filepath)
datafile = Dataset_File(dataset=dataset,
filename=path.basename(filepath),
size=size)
replica = Replica(datafile=datafile,
protocol='staging',
url=url,
location=Location.get_location('staging'))
replica.verify(allowEmptyChecksums=True)
datafile.save()
replica.datafile = datafile
replica.save()
def testMirror(self):
dest = Location.get_location('test')
datafile, replica = generate_datafile(None, self.dataset, "Hi granny")
path = datafile.get_absolute_filepath()
self.assertTrue(os.path.exists(path))
dummy_replica = Replica()
dummy_replica.datafile = datafile
dummy_replica.location = Location.objects.get(name='test')
dummy_replica.url = dummy_replica.generate_default_url()
with self.assertRaises(TransferError):
dest.provider.get_length(dummy_replica)
self.assertTrue(migrate_replica(replica, dest, mirror=True))
datafile = Dataset_File.objects.get(id=datafile.id)
self.assertTrue(datafile.is_local())
self.assertEquals(dest.provider.get_length(dummy_replica), 9)
def fpupload(request, dataset_id):
"""
Uploads all files picked by filepicker to the dataset
:param request: a HTTP Request instance
:type request: :class:`django.http.HttpRequest`
:param dataset_id: the dataset_id
:type dataset_id: integer
:returns: boolean true if successful
:rtype: bool
"""
dataset = Dataset.objects.get(id=dataset_id)
logger.debug('called fpupload')
if request.method == 'POST':
logger.debug('got POST')
for key, val in request.POST.items():
splits = val.split(",")
for url in splits:
try:
fp = FilepickerFile(url)
except ValueError:
pass
else:
picked_file = fp.get_file()
filepath = write_uploaded_file_to_dataset(
dataset, picked_file)
datafile = Dataset_File(dataset=dataset,
filename=picked_file.name,
size=picked_file.size)
replica = Replica(datafile=datafile,
url=filepath,
protocol='',
location=Location.get_default_location())
replica.verify(allowEmptyChecksums=True)
datafile.save()
replica.datafile = datafile
replica.save()
return HttpResponse(json.dumps({"result": True}))
def migrate_replica(replica, location, noRemove=False, mirror=False):
"""
Migrate the replica to a different storage location. The overall
effect will be that the datafile will be stored at the new location and
removed from the current location, and the datafile metadata will be
updated to reflect this.
"""
from tardis.tardis_portal.models import Replica, Location
with transaction.commit_on_success():
replica = Replica.objects.select_for_update().get(pk=replica.pk)
source = Location.get_location(replica.location.name)
if not replica.verified or location.provider.trust_length:
raise MigrationError('Only verified datafiles can be migrated' \
' to this destination')
filename = replica.get_absolute_filepath()
try:
newreplica = Replica.objects.get(datafile=replica.datafile,
location=location)
created_replica = False
# We've most likely mirrored this file previously. But if
# we are about to delete the source Replica, we need to check
# that the target Replica still verifies.
if not mirror and not check_file_transferred(newreplica, location):
raise MigrationError('Previously mirrored / migrated Replica' \
' no longer verifies locally!')
except Replica.DoesNotExist:
newreplica = Replica()
newreplica.location = location
newreplica.datafile = replica.datafile
newreplica.protocol = ''
newreplica.stay_remote = location != Location.get_default_location(
)
newreplica.verified = False
url = location.provider.generate_url(newreplica)
if newreplica.url == url:
# We should get here ...
raise MigrationError('Cannot migrate a replica to its' \
' current location')
newreplica.url = url
location.provider.put_file(replica, newreplica)
verified = False
try:
verified = check_file_transferred(newreplica, location)
except:
# FIXME - should we always do this?
location.provider.remove_file(newreplica)
raise
newreplica.verified = verified
newreplica.save()
logger.info('Transferred file %s for replica %s' %
(filename, replica.id))
created_replica = True
if mirror:
return created_replica
# FIXME - do this more reliably ...
replica.delete()
if not noRemove:
source.provider.remove_file(replica)
logger.info('Removed local file %s for replica %s' %
(filename, replica.id))
return True