def test_save_metadata(self):
"""
Tests that the metadata XML is not written to the storage folder for internal
resources but only when the resource is ingested
"""
# load test fixture; its initial status is 'internal'
_result = test_utils.import_xml(TESTFIXTURE_XML)
resource = resourceInfoType_model.objects.get(pk=_result.id)
_storage_object = resource.storage_object
_storage_object.update_storage()
# initial status is 'internal'
self.assertTrue(_storage_object.publication_status == INTERNAL)
# internal resource has no metadata XML stored in storage folder
self.assertFalse(
os.path.isfile('{0}/metadata-{1:04d}.xml'.format(
_storage_object._storage_folder(), _storage_object.revision)))
# set status to ingested
_storage_object.publication_status = INGESTED
_storage_object.update_storage()
# ingested resource has metadata XML stored in storage folder
self.assertTrue(
os.path.isfile('{0}/metadata-{1:04d}.xml'.format(
_storage_object._storage_folder(), _storage_object.revision)))
# ingested resource has global part of storage object in storage folder
self.assertTrue(
os.path.isfile('{0}/storage-global.json'.format(
_storage_object._storage_folder())))
# ingested resource has local part of storage object in storage folder
self.assertTrue(
os.path.isfile('{0}/storage-local.json'.format(
_storage_object._storage_folder())))
# ingested resource has digest zip in storage folder
self.assertTrue(
os.path.isfile('{0}/resource.zip'.format(
_storage_object._storage_folder())))
# digest zip contains metadata.xml and storage-global.json
_zf_name = '{0}/resource.zip'.format( _storage_object._storage_folder())
_zf = zipfile.ZipFile(_zf_name, mode='r')
self.assertTrue('metadata.xml' in _zf.namelist())
self.assertTrue('storage-global.json' in _zf.namelist())
# md5 of digest zip is stored in storage object
with ZipFile(_zf_name, 'r') as inzip:
with inzip.open('metadata.xml') as resource_xml:
resource_xml_string = resource_xml.read()
with inzip.open('storage-global.json') as storage_file:
# read json string
storage_json_string = storage_file.read()
_checksum = compute_digest_checksum(
resource_xml_string, storage_json_string)
self.assertEqual(_checksum, _storage_object.digest_checksum)
def test_metadata_digest(self):
settings.SYNC_NEEDS_AUTHENTICATION = False
client = Client()
resource = resourceInfoType_model.objects.all()[0]
resource_uuid = resource.storage_object.identifier
expected_digest = resource.storage_object.digest_checksum
response = client.get("{0}{1}/metadata/".format(self.SYNC_BASE, resource_uuid))
# read zip file from response
with ZipFile(StringIO(response.content), "r") as inzip:
with inzip.open("metadata.xml") as resource_xml:
resource_xml_string = resource_xml.read()
with inzip.open("storage-global.json") as storage_file:
# should be a json object, not string
storage_json_string = storage_file.read()
self.assertEquals(expected_digest, compute_digest_checksum(resource_xml_string, storage_json_string))
def test_metadata_digest(self):
settings.SYNC_NEEDS_AUTHENTICATION = False
client = Client()
resource = resourceInfoType_model.objects.all()[0]
resource_uuid = resource.storage_object.identifier
expected_digest = resource.storage_object.digest_checksum
response = client.get('{0}{1}/metadata/'.format(
self.SYNC_BASE, resource_uuid))
# read zip file from response
with ZipFile(StringIO(response.content), 'r') as inzip:
with inzip.open('metadata.xml') as resource_xml:
resource_xml_string = resource_xml.read()
with inzip.open('storage-global.json') as storage_file:
# should be a json object, not string
storage_json_string = storage_file.read()
self.assertEquals(
expected_digest,
compute_digest_checksum(resource_xml_string, storage_json_string))
def get_full_metadata(opener, full_metadata_url, expected_digest):
"""
Obtain the full metadata record for one resource.
Returns a pair of storage_json_string, resource_xml_string.
Raises CorruptDataException if the zip data received from full_metadata_url
does not have an md5 digest identical to expected_digest.
"""
with contextlib.closing(opener.open(full_metadata_url)) as response:
data = response.read()
with ZipFile(StringIO(data), "r") as inzip:
with inzip.open("metadata.xml") as resource_xml:
resource_xml_string = resource_xml.read()
with inzip.open("storage-global.json") as storage_file:
# read json string
storage_json_string = storage_file.read()
# convert to json object
storage_json = json.loads(storage_json_string)
if not expected_digest == compute_digest_checksum(resource_xml_string, storage_json_string):
raise CorruptDataException("Checksum error for resource '{0}'.".format(full_metadata_url))
return storage_json, resource_xml_string
def get_full_metadata(opener, full_metadata_url, expected_digest):
"""
Obtain the full metadata record for one resource.
Returns a pair of storage_json_string, resource_xml_string.
Raises CorruptDataException if the zip data received from full_metadata_url
does not have an md5 digest identical to expected_digest.
"""
with contextlib.closing(opener.open(full_metadata_url)) as response:
data = response.read()
with ZipFile(StringIO(data), 'r') as inzip:
with inzip.open('metadata.xml') as resource_xml:
resource_xml_string = resource_xml.read()
with inzip.open('storage-global.json') as storage_file:
# read json string
storage_json_string = storage_file.read()
# convert to json object
storage_json = json.loads(storage_json_string)
if not expected_digest == \
compute_digest_checksum(resource_xml_string, storage_json_string):
raise CorruptDataException("Checksum error for resource '{0}'." \
.format(full_metadata_url))
return storage_json, resource_xml_string
