def test_get_values_for_attribute_3(self):
raw_meta = IrodsRawFileMetadata(fpath='/seq/123/123.bam')
raw_meta.init_avus_from_avu_tuples([MetaAVU(attribute='sample', value='ABCSample'),
MetaAVU(attribute='library', value='lib')])
expected_result = set()
actual_result = raw_meta.get_values_for_attribute('study')
self.assertSetEqual(actual_result, expected_result)
def test_check_more_than_one_replicas_when_1_replica(self):
replicas = [
baton_models.DataObjectReplica(number=1, checksum="123abc")]
raw_metadata = IrodsRawFileMetadata(fpath='/seq/123/123.cram', file_replicas=replicas)
result = raw_metadata.ReplicasChecks.check_more_than_one_replicas(raw_metadata.file_replicas)
#self.assertEqual(len(result), 1)
self.assertEqual(result.result, RESULT.FAILURE)
def test_from_baton_wrapper_file_replicas(self):
replicas = [
baton_models.DataObjectReplica(number=1, checksum="123abc"),
baton_models.DataObjectReplica(number=2, checksum="abc"),]
data_obj = baton_models.DataObject(path='/humgen/projects/helic/123.bam', replicas=replicas)
raw_meta = IrodsRawFileMetadata.from_baton_wrapper(data_obj)
self.assertEqual(len(raw_meta.file_replicas), 2)
def retrieve_raw_files_metadata_by_metadata(cls,
search_criteria_list: List[
Tuple],
zone=None):
search_criteria_objs = []
for k, v in search_criteria_list:
search_criterion = SearchCriterion(k, v)
search_criteria_objs.append(search_criterion)
# Getting metadata from iRODS:
try:
connection = connect_to_irods_with_baton(
config.BATON_BIN
) # skip_baton_binaries_validation=True) # type: Connection
list_of_data_objs_and_metadata = connection.data_object.get_by_metadata(
search_criteria_objs, zone=zone)
except RuntimeError as e:
if str(e).find('KRB_ERROR_ACQUIRING_CREDS') != -1:
raise OSError(
"ERROR: you need to log into iRODS and aquire the KERBEROS credentials."
) from None
else:
raise e from None
raw_meta_objects = [
IrodsRawFileMetadata.from_baton_wrapper(data_obj)
for data_obj in list_of_data_objs_and_metadata
]
return raw_meta_objects
def test_from_baton_wrapper_missing_bits(self):
replicas = [
baton_models.DataObjectReplica(number=1, checksum="123abc", host='hgi-dev', resource_name='irods-s1', up_to_date=True)]
metadata = baton_coll.IrodsMetadata({'study': set(['BLUEPRINT'])})
data_obj = baton_models.DataObject(path='/somepath/file.txt', metadata=metadata, replicas=replicas)
raw_meta = IrodsRawFileMetadata.from_baton_wrapper(data_obj)
self.assertEqual(len(raw_meta.file_replicas), 1)
self.assertEqual(len(raw_meta.acls), 0)
def test_from_baton_wrapper_acls(self):
user = "******"
acl = [baton_models.AccessControl(user, level=baton_models.AccessControl.Level.WRITE)]
data_obj = baton_models.DataObject(path='/somepath/file.txt', access_controls=acl)
raw_meta = IrodsRawFileMetadata.from_baton_wrapper(data_obj)
self.assertEqual(len(raw_meta.acls), 1)
self.assertEqual(raw_meta.acls[0].access_group, 'hgi')
self.assertEqual(raw_meta.acls[0].permission, IrodsPermission.WRITE)
self.assertEqual(raw_meta.acls[0].zone, 'Sanger1')
def test_from_raw_metadata_only_replicas(self):
replicas = [
baton_models.DataObjectReplica(number=1, checksum="123abc"),
baton_models.DataObjectReplica(number=2, checksum="abc"),]
raw_metadata = IrodsRawFileMetadata(fpath='/seq/123.bam', file_replicas=replicas)
seq_metadata = IrodsSeqFileMetadata.from_raw_metadata(raw_metadata)
expected = {'name': set(), 'accession_number': set(), 'internal_id': set()}
self.assertEqual(seq_metadata.samples, expected)
self.assertEqual(seq_metadata.libraries, expected)
self.assertEqual(seq_metadata.checksum_in_meta, set())
def test_check_has_read_permission_ss_group_no_ss_grp(self):
acl1 = IrodsACL(access_group='public#seq', zone='seq', permission='read')
acl2 = IrodsACL(access_group='npg#seq', zone='seq', permission='own')
raw_metadata = IrodsRawFileMetadata(fpath='/home/myfile', acls=[acl1, acl2])
result = raw_metadata.ACLsChecks.check_read_permission_exists_for_ss_group(raw_metadata.acls)
self.assertEqual(len(result), 2)
for check_res in result:
if check_res.check_name in [CHECK_NAMES.check_ss_irods_group_read_permission, CHECK_NAMES.check_there_is_ss_irods_group]:
self.assertEqual(check_res.result, RESULT.FAILURE)
else:
self.assertEqual(check_res.result, RESULT.SUCCESS)
def test_from_baton_wrapper_all_ok(self):
user = "******"
acl = [baton_models.AccessControl(user, level=baton_models.AccessControl.Level.OWN)]
replicas = [
baton_models.DataObjectReplica(number=1, checksum="123abc", host='hgi-dev', resource_name='irods-s1', up_to_date=True)]
data_obj = baton_models.DataObject(path='/somepath/file.txt', access_controls=acl, replicas=replicas)
raw_meta = IrodsRawFileMetadata.from_baton_wrapper(data_obj)
self.assertEqual(raw_meta.fpath, '/somepath/file.txt')
self.assertEqual(len(raw_meta.file_replicas), 1)
self.assertEqual(len(raw_meta.acls), 1)
self.assertEqual(raw_meta.acls[0].zone, 'humgen')
self.assertEqual(raw_meta.acls[0].access_group, 'serapis')
def fetch_raw_file_metadata_by_path(cls, fpath):
try:
connection = connect_to_irods_with_baton(config.BATON_BIN)#, skip_baton_binaries_validation=True)
data_object = connection.data_object.get_by_path(fpath)
except Exception as e:
if str(e).find('KRB_ERROR_ACQUIRING_CREDS') != -1:
raise OSError("ERROR: you need to log into iRODS and aquire the KERBEROS credentials.") from None
else:
raise e from None
else:
if data_object:
raw_metadata = IrodsRawFileMetadata.from_baton_wrapper(data_object)
return raw_metadata
return None
def test_from_baton_wrapper_full_obj(self):
user = "******"
acl = [baton_models.AccessControl(user, level=baton_models.AccessControl.Level.OWN)]
replicas = [
baton_models.DataObjectReplica(number=1, checksum="123abc", host='hgi-dev', resource_name='irods-s1', up_to_date=True),
baton_models.DataObjectReplica(number=2, checksum="abc", host='hgi-dev-wow', resource_name='irods-s2', up_to_date=True),]
metadata = baton_coll.IrodsMetadata({'study': set(['BLUEPRINT'])})
data_obj = baton_models.DataObject(path='/somepath/file.txt', access_controls=acl, metadata=metadata, replicas=replicas)
raw_meta = IrodsRawFileMetadata.from_baton_wrapper(data_obj)
self.assertEqual(raw_meta.fpath, '/somepath/file.txt')
self.assertEqual(len(raw_meta.file_replicas), 2)
self.assertEqual(len(raw_meta.acls), 1)
self.assertEqual(raw_meta.acls[0].zone, 'humgen')
self.assertEqual(raw_meta.acls[0].access_group, 'hgi')
def retrieve_raw_files_metadata_by_metadata(cls, search_criteria_list: List[Tuple], zone=None):
search_criteria_objs = []
for k, v in search_criteria_list:
search_criterion = SearchCriterion(k, v)
search_criteria_objs.append(search_criterion)
# Getting metadata from iRODS:
try:
connection = connect_to_irods_with_baton(config.BATON_BIN) # skip_baton_binaries_validation=True) # type: Connection
list_of_data_objs_and_metadata = connection.data_object.get_by_metadata(search_criteria_objs, zone=zone)
except RuntimeError as e:
if str(e).find('KRB_ERROR_ACQUIRING_CREDS') != -1:
raise OSError("ERROR: you need to log into iRODS and aquire the KERBEROS credentials.") from None
else:
raise e from None
raw_meta_objects = [IrodsRawFileMetadata.from_baton_wrapper(data_obj) for data_obj in list_of_data_objs_and_metadata]
return raw_meta_objects
def fetch_raw_file_metadata_by_path(cls, fpath):
try:
connection = connect_to_irods_with_baton(
config.BATON_BIN) #, skip_baton_binaries_validation=True)
data_object = connection.data_object.get_by_path(fpath)
except Exception as e:
if str(e).find('KRB_ERROR_ACQUIRING_CREDS') != -1:
raise OSError(
"ERROR: you need to log into iRODS and aquire the KERBEROS credentials."
) from None
else:
raise e from None
else:
if data_object:
raw_metadata = IrodsRawFileMetadata.from_baton_wrapper(
data_object)
return raw_metadata
return None
def test_from_baton_wrapper_fpath_1(self):
data_obj = baton_models.DataObject(path='/seq/123/123.bam.bai')
raw_meta = IrodsRawFileMetadata.from_baton_wrapper(data_obj)
self.assertEqual(raw_meta.fpath, '/seq/123/123.bam.bai')
def test_from_baton_wrapper_metadata(self):
metadata = baton_coll.IrodsMetadata({'study': set(['BLUEPRINT']), 'sample': set(['123sam'])})
data_obj = baton_models.DataObject(path='/somepath/file.txt', metadata=metadata)
raw_meta = IrodsRawFileMetadata.from_baton_wrapper(data_obj)
self.assertEqual(raw_meta.get_values_for_attribute('study'), set(['BLUEPRINT']))
def test_set_attributes_from_avus_1(self):
raw_meta = IrodsRawFileMetadata(fpath='/seq/123/123.bam')
raw_meta.init_avus_from_avu_tuples([MetaAVU(attribute='sample', value='ABCSample'),
MetaAVU(attribute='library', value='lib')])
expected_result = {'sample': set(['ABCSample']), 'library': set(['lib'])}
self.assertDictEqual(raw_meta.avus, expected_result)
def test_get_values_count_for_attribute_1(self):
raw_meta = IrodsRawFileMetadata(fpath='/seq/123/123.bam')
raw_meta.init_avus_from_avu_tuples([MetaAVU(attribute='sample', value='ABCSample')])
expected_result = 1
actual_result = raw_meta.get_values_count_for_attribute('sample')
self.assertEqual(expected_result, actual_result)
def test_get_values_count_for_attribute_2(self):
raw_meta = IrodsRawFileMetadata(fpath='/seq/123/123.bam')
expected_result = 0
actual_result = raw_meta.get_values_count_for_attribute('sample')
self.assertEqual(expected_result, actual_result)
def test_check_more_than_one_replicas_when_no_replica(self):
replicas = []
raw_metadata = IrodsRawFileMetadata(fpath='/seq/123/123.cram', file_replicas=replicas)
result = raw_metadata.ReplicasChecks.check_more_than_one_replicas(raw_metadata.file_replicas)
#self.assertEqual(len(result), 1)
self.assertEqual(result.result, RESULT.FAILURE)
def test_is_true_comparison_when_less(self):
result = IrodsRawFileMetadata._is_true_comparison(1, 2, '<')
self.assertTrue(result)
def test_is_true_comparison_when_not_equal(self):
result = IrodsRawFileMetadata._is_true_comparison(1, 3, '=')
self.assertFalse(result)
def test_from_baton_wrapper_fpath_2(self):
data_obj = baton_models.DataObject(path='/humgen/projects/helic/123.bam')
raw_meta = IrodsRawFileMetadata.from_baton_wrapper(data_obj)
self.assertEqual(raw_meta.fpath, '/humgen/projects/helic/123.bam')
def test_check_attribute_count_equal_ok(self):
raw_metadata = IrodsRawFileMetadata(fpath='/home/myfile')
raw_metadata.avus = {'sample': set(['1', '2', '3'])}
result = raw_metadata.check_attribute_count([AttributeCount(attribute='sample', count=3, operator='=')])
#self.assertEqual(len(result), 1)
self.assertEqual(result.result, RESULT.SUCCESS)
def test_check_attribute_count_when_not_found(self):
raw_metadata = IrodsRawFileMetadata(fpath='/home/myfile')
raw_metadata.avus = {'sample': set(['1', '2', '3'])}
result = raw_metadata.check_attribute_count([AttributeCount(attribute='study', count=2, operator='>')])
#self.assertEqual(len(result), 1)
self.assertEqual(result.result, RESULT.FAILURE)
