def setUp(self):
super(LargeFileTest, self).setUp()
url = self._get_account_url()
self.payload_dropping_policy = PayloadDroppingPolicy()
credential_policy = _format_shared_key_credential(
self.settings.STORAGE_DATA_LAKE_ACCOUNT_NAME,
self.settings.STORAGE_DATA_LAKE_ACCOUNT_KEY)
self.dsc = DataLakeServiceClient(
url,
credential=self.settings.STORAGE_DATA_LAKE_ACCOUNT_KEY,
_additional_pipeline_policies=[
self.payload_dropping_policy, credential_policy
])
loop = asyncio.get_event_loop()
loop.run_until_complete(self.dsc.__aenter__())
self.config = self.dsc._config
self.file_system_name = self.get_resource_name('filesystem')
if not self.is_playback():
file_system = self.dsc.get_file_system_client(
self.file_system_name)
try:
loop = asyncio.get_event_loop()
loop.run_until_complete(
file_system.create_file_system(timeout=5))
except ResourceExistsError:
pass
def _setUp(self, account_name, account_key):
url = self.account_url(account_name, 'dfs')
self.dsc = DataLakeServiceClient(url,
credential=account_key,
transport=AiohttpTestTransport(),
logging_enable=True)
self.config = self.dsc._config
self.test_file_systems = []
def setUp(self):
super(FileSystemTest, self).setUp()
url = self._get_account_url()
self.dsc = DataLakeServiceClient(url, credential=self.settings.STORAGE_DATA_LAKE_ACCOUNT_KEY,
transport=AiohttpTestTransport())
self.config = self.dsc._config
loop = asyncio.get_event_loop()
loop.run_until_complete(self.dsc.__aenter__())
self.test_file_systems = []
async def test_service_client_session_closes_after_filesystem_creation(
self, datalake_storage_account_name, datalake_storage_account_key):
self._setUp(datalake_storage_account_name, datalake_storage_account_key)
# Arrange
dsc2 = DataLakeServiceClient(self.dsc.url, credential=datalake_storage_account_key)
async with DataLakeServiceClient(
self.dsc.url, credential=datalake_storage_account_key) as ds_client:
fs1 = await ds_client.create_file_system(self._get_file_system_reference(prefix="fs1"))
await fs1.delete_file_system()
await dsc2.create_file_system(self._get_file_system_reference(prefix="fs2"))
await dsc2.close()
async def _setUp(self, account_name, account_key):
url = self.account_url(account_name, 'dfs')
self.dsc = DataLakeServiceClient(url, credential=account_key)
self.config = self.dsc._config
self.file_system_name = self.get_resource_name('filesystem')
if not self.is_playback():
file_system = self.dsc.get_file_system_client(self.file_system_name)
try:
await file_system.create_file_system(timeout=5)
except ResourceExistsError:
pass
async def _test_read_file_with_user_delegation_key(self):
# SAS URL is calculated from storage key, so this test runs live only
if TestMode.need_recording_file(self.test_mode):
return
# Create file
file_client = await self._create_file_and_return_client()
data = self.get_random_bytes(1024)
# Upload data to file
await file_client.append_data(data, 0, len(data))
await file_client.flush_data(len(data))
# Get user delegation key
token_credential = self.generate_async_oauth_token()
service_client = DataLakeServiceClient(self._get_oauth_account_url(), credential=token_credential)
user_delegation_key = await service_client.get_user_delegation_key(datetime.utcnow(),
datetime.utcnow() + timedelta(hours=1))
sas_token = generate_file_sas(file_client.account_name,
file_client.file_system_name,
None,
file_client.path_name,
user_delegation_key=user_delegation_key,
permission=FileSasPermissions(read=True, create=True, write=True, delete=True),
expiry=datetime.utcnow() + timedelta(hours=1),
)
# doanload the data and make sure it is the same as uploaded data
new_file_client = DataLakeFileClient(self._get_account_url(),
file_client.file_system_name,
file_client.path_name,
credential=sas_token)
downloaded_data = await new_file_client.read_file()
self.assertEqual(data, downloaded_data)
async def list_paths_in_file_system(self):
# Instantiate a DataLakeServiceClient using a connection string
from azure.storage.filedatalake.aio import DataLakeServiceClient
datalake_service_client = DataLakeServiceClient.from_connection_string(
self.connection_string)
async with datalake_service_client:
# Instantiate a FileSystemClient
file_system_client = datalake_service_client.get_file_system_client(
"mypathfilesystem")
# Create new File System
await file_system_client.create_file_system()
# [START upload_file_to_file_system]
file_client = file_system_client.get_file_client("myfile")
await file_client.create_file()
with open(SOURCE_FILE, "rb") as data:
length = data.tell()
await file_client.append_data(data, 0)
await file_client.flush_data(length)
# [END upload_file_to_file_system]
# [START get_paths_in_file_system]
path_list = file_system_client.get_paths()
async for path in path_list:
print(path.name + '\n')
# [END get_paths_in_file_system]
# Delete file system
await file_system_client.delete_file_system()
async def test_restore_file_system_with_sas(self, datalake_storage_account_name, datalake_storage_account_key):
pytest.skip(
"We are generating a SAS token therefore play only live but we also need a soft delete enabled account.")
self._setUp(datalake_storage_account_name, datalake_storage_account_key)
token = generate_account_sas(
self.dsc.account_name,
self.dsc.credential.account_key,
ResourceTypes(service=True, file_system=True),
AccountSasPermissions(read=True, write=True, list=True, delete=True),
datetime.utcnow() + timedelta(hours=1),
)
dsc = DataLakeServiceClient(self.dsc.url, token)
name = self._get_file_system_reference(prefix="filesystem")
filesystem_client = await dsc.create_file_system(name)
await filesystem_client.delete_file_system()
# to make sure the filesystem is deleted
with self.assertRaises(ResourceNotFoundError):
await filesystem_client.get_file_system_properties()
filesystem_list = []
async for fs in self.dsc.list_file_systems(include_deleted=True):
filesystem_list.append(fs)
self.assertTrue(len(filesystem_list) >= 1)
restored_version = 0
for filesystem in filesystem_list:
# find the deleted filesystem and restore it
if filesystem.deleted and filesystem.name == filesystem_client.file_system_name:
restored_fs_client = await dsc.undelete_file_system(filesystem.name, filesystem.deleted_version,
new_name="restored" + name + str(restored_version))
restored_version += 1
# to make sure the deleted filesystem is restored
props = await restored_fs_client.get_file_system_properties()
self.assertIsNotNone(props)
async def test_get_access_control_using_delegation_sas_async(
self, datalake_storage_account_name, datalake_storage_account_key):
self._setUp(datalake_storage_account_name, datalake_storage_account_key)
url = self._get_account_url(datalake_storage_account_name)
token_credential = self.generate_oauth_token()
dsc = DataLakeServiceClient(url, token_credential, logging_enable=True)
file_system_name = self._get_file_system_reference()
directory_client_name = '/'
(await dsc.create_file_system(file_system_name)).get_directory_client(directory_client_name)
directory_client = self.dsc.get_directory_client(file_system_name, directory_client_name)
random_guid = uuid.uuid4()
await directory_client.set_access_control(owner=random_guid,
permissions='0777')
acl = await directory_client.get_access_control()
delegation_key = await dsc.get_user_delegation_key(datetime.utcnow(),
datetime.utcnow() + timedelta(hours=1))
token = generate_file_system_sas(
dsc.account_name,
file_system_name,
delegation_key,
permission=FileSystemSasPermissions(
read=True, execute=True, manage_access_control=True, manage_ownership=True),
expiry=datetime.utcnow() + timedelta(hours=1),
agent_object_id=random_guid
)
sas_directory_client = DataLakeDirectoryClient(self.dsc.url, file_system_name, directory_client_name,
credential=token, logging_enable=True)
access_control = await sas_directory_client.get_access_control()
self.assertIsNotNone(access_control)
async def _test_service_client_session_closes_after_filesystem_creation(
self):
# Arrange
dsc2 = DataLakeServiceClient(
self.dsc.url,
credential=self.settings.STORAGE_DATA_LAKE_ACCOUNT_KEY)
async with DataLakeServiceClient(
self.dsc.url,
credential=self.settings.STORAGE_DATA_LAKE_ACCOUNT_KEY
) as ds_client:
fs1 = await ds_client.create_file_system(
self._get_file_system_reference(prefix="fs1"))
await fs1.delete_file_system()
await dsc2.create_file_system(
self._get_file_system_reference(prefix="fs2"))
await dsc2.close()
async def test_list_paths_using_file_sys_delegation_sas_async(
self, datalake_storage_account_name, datalake_storage_account_key):
self._setUp(datalake_storage_account_name, datalake_storage_account_key)
url = self._get_account_url(datalake_storage_account_name)
token_credential = self.generate_oauth_token()
dsc = DataLakeServiceClient(url, token_credential)
file_system_name = self._get_file_system_reference()
directory_client_name = '/'
directory_client = (await dsc.create_file_system(file_system_name)).get_directory_client(directory_client_name)
random_guid = uuid.uuid4()
await directory_client.set_access_control(owner=random_guid, permissions='0777')
delegation_key = await dsc.get_user_delegation_key(datetime.utcnow(),
datetime.utcnow() + timedelta(hours=1))
token = generate_file_system_sas(
dsc.account_name,
file_system_name,
delegation_key,
permission=DirectorySasPermissions(list=True),
expiry=datetime.utcnow() + timedelta(hours=1),
agent_object_id=random_guid
)
sas_directory_client = FileSystemClient(self.dsc.url, file_system_name,
credential=token)
paths = list()
async for path in sas_directory_client.get_paths():
paths.append(path)
self.assertEqual(0, 0)
async def set_metadata_on_file_system(self):
# Instantiate a DataLakeServiceClient using a connection string
from azure.storage.filedatalake.aio import DataLakeServiceClient
datalake_service_client = DataLakeServiceClient.from_connection_string(
self.connection_string)
async with datalake_service_client:
# Instantiate a FileSystemClient
file_system_client = datalake_service_client.get_file_system_client(
"mymetadatafilesystemsync")
try:
# Create new File System
await file_system_client.create_file_system()
# [START set_file_system_metadata]
# Create key, value pairs for metadata
metadata = {'type': 'test'}
# Set metadata on the file system
await file_system_client.set_file_system_metadata(
metadata=metadata)
# [END set_file_system_metadata]
# Get file system properties
properties = await file_system_client.get_file_system_properties(
)
finally:
# Delete file system
await file_system_client.delete_file_system()
async def acquire_lease_on_file_system(self):
# Instantiate a DataLakeServiceClient using a connection string
# [START create_data_lake_service_client_from_conn_str]
from azure.storage.filedatalake.aio import DataLakeServiceClient
datalake_service_client = DataLakeServiceClient.from_connection_string(
self.connection_string)
# [END create_data_lake_service_client_from_conn_str]
async with datalake_service_client:
# Instantiate a FileSystemClient
file_system_client = datalake_service_client.get_file_system_client(
"myleasefilesystem")
# Create new File System
try:
await file_system_client.create_file_system()
except ResourceExistsError:
pass
# [START acquire_lease_on_file_system]
# Acquire a lease on the file system
lease = await file_system_client.acquire_lease()
# Delete file system by passing in the lease
await file_system_client.delete_file_system(lease=lease)
async def file_system_sample(self):
# [START create_file_system_client_from_service]
# Instantiate a DataLakeServiceClient using a connection string
from azure.storage.filedatalake.aio import DataLakeServiceClient
datalake_service_client = DataLakeServiceClient.from_connection_string(
self.connection_string)
async with datalake_service_client:
# Instantiate a FileSystemClient
file_system_client = datalake_service_client.get_file_system_client(
"mynewfilesystems")
# [END create_file_system_client_from_service]
try:
# [START create_file_system]
await file_system_client.create_file_system()
# [END create_file_system]
# [START get_file_system_properties]
properties = await file_system_client.get_file_system_properties(
)
# [END get_file_system_properties]
finally:
# [START delete_file_system]
await file_system_client.delete_file_system()
async def get_directory_client_from_file_system(self):
# Instantiate a DataLakeServiceClient using a connection string
from azure.storage.filedatalake.aio import DataLakeServiceClient
datalake_service_client = DataLakeServiceClient.from_connection_string(
self.connection_string)
async with datalake_service_client:
# Instantiate a FileSystemClient
file_system_client = datalake_service_client.get_file_system_client(
"mydirectoryfilesystem")
# Create new File System
try:
await file_system_client.create_file_system()
except ResourceExistsError:
pass
# [START get_directory_client_from_file_system]
# Get the DataLakeDirectoryClient from the FileSystemClient to interact with a specific file
directory_client = file_system_client.get_directory_client(
"mynewdirectory")
# [END get_directory_client_from_file_system]
# Delete file system
await file_system_client.delete_file_system()
async def run():
account_name = os.getenv('STORAGE_ACCOUNT_NAME', "")
account_key = os.getenv('STORAGE_ACCOUNT_KEY', "")
# set up the service client with the credentials from the environment variables
service_client = DataLakeServiceClient(
account_url="{}://{}.dfs.core.windows.net".format(
"https", account_name),
credential=account_key)
async with service_client:
# generate a random name for testing purpose
fs_name = "testfs{}".format(random.randint(1, 1000))
print("Generating a test filesystem named '{}'.".format(fs_name))
# create the filesystem
filesystem_client = await service_client.create_file_system(
file_system=fs_name)
# invoke the sample code
try:
await directory_sample(filesystem_client)
finally:
# clean up the demo filesystem
await filesystem_client.delete_file_system()
def setUp(self):
super(DirectoryTest, self).setUp()
url = self._get_account_url()
self.dsc = DataLakeServiceClient(url, credential=self.settings.STORAGE_DATA_LAKE_ACCOUNT_KEY,
transport=AiohttpTestTransport())
loop = asyncio.get_event_loop()
loop.run_until_complete(self.dsc.__aenter__())
self.config = self.dsc._config
self.file_system_name = self.get_resource_name('filesystem')
if not self.is_playback():
file_system = self.dsc.get_file_system_client(self.file_system_name)
try:
loop.run_until_complete(file_system.create_file_system(timeout=5))
except ResourceExistsError:
pass
async def test_list_system_filesystems_async(self,
datalake_storage_account_name,
datalake_storage_account_key):
self._setUp(datalake_storage_account_name,
datalake_storage_account_key)
# Arrange
dsc = DataLakeServiceClient(self.dsc.url,
credential=datalake_storage_account_key)
# Act
filesystems = []
async for fs in dsc.list_file_systems(include_system=True):
filesystems.append(fs)
# Assert
found = False
for fs in filesystems:
if fs.name == "$logs":
found = True
self.assertEqual(found, True)
async def _setUp(self, account_name, account_key):
url = self._get_account_url(account_name)
self.payload_dropping_policy = PayloadDroppingPolicy()
credential_policy = _format_shared_key_credential(account_name, account_key)
self.dsc = DataLakeServiceClient(url,
credential=account_key,
_additional_pipeline_policies=[self.payload_dropping_policy, credential_policy])
self.config = self.dsc._config
self.file_system_name = self.get_resource_name('filesystem')
if not self.is_playback():
file_system = self.dsc.get_file_system_client(self.file_system_name)
try:
await file_system.create_file_system(timeout=5)
except ResourceExistsError:
pass
class FileTest(StorageTestCase):
def setUp(self):
super(FileTest, self).setUp()
url = self._get_account_url()
self.dsc = DataLakeServiceClient(
url, credential=self.settings.STORAGE_DATA_LAKE_ACCOUNT_KEY)
loop = asyncio.get_event_loop()
loop.run_until_complete(self.dsc.__aenter__())
self.config = self.dsc._config
self.file_system_name = self.get_resource_name('filesystem')
if not self.is_playback():
file_system = self.dsc.get_file_system_client(
self.file_system_name)
try:
loop = asyncio.get_event_loop()
loop.run_until_complete(
file_system.create_file_system(timeout=5))
except ResourceExistsError:
pass
def tearDown(self):
if not self.is_playback():
try:
loop = asyncio.get_event_loop()
loop.run_until_complete(
self.dsc.delete_file_system(self.file_system_name))
loop.run_until_complete(self.dsc.__aexit__())
except:
pass
return super(FileTest, self).tearDown()
# --Helpers-----------------------------------------------------------------
def _get_directory_reference(self, prefix=TEST_DIRECTORY_PREFIX):
directory_name = self.get_resource_name(prefix)
return directory_name
def _get_file_reference(self, prefix=TEST_FILE_PREFIX):
file_name = self.get_resource_name(prefix)
return file_name
async def _create_file_system(self):
return await self.dsc.create_file_system(
self._get_file_system_reference())
async def _create_directory_and_return_client(self, directory=None):
directory_name = directory if directory else self._get_directory_reference(
)
directory_client = self.dsc.get_directory_client(
self.file_system_name, directory_name)
await directory_client.create_directory()
return directory_client
async def _create_file_and_return_client(self, directory="", file=None):
if directory:
await self._create_directory_and_return_client(directory)
if not file:
file = self._get_file_reference()
file_client = self.dsc.get_file_client(self.file_system_name,
directory + '/' + file)
await file_client.create_file()
return file_client
# --Helpers-----------------------------------------------------------------
async def _test_create_file(self):
# Arrange
directory_name = self._get_directory_reference()
# Create a directory to put the file under that
directory_client = self.dsc.get_directory_client(
self.file_system_name, directory_name)
await directory_client.create_directory()
file_client = directory_client.get_file_client('filename')
response = await file_client.create_file()
# Assert
self.assertIsNotNone(response)
@record
def test_create_file_async(self):
loop = asyncio.get_event_loop()
loop.run_until_complete(self._test_create_file())
async def _test_create_file_using_oauth_token_credential(self):
# Arrange
file_name = self._get_file_reference()
token_credential = self.generate_async_oauth_token()
# Create a directory to put the file under that
file_client = DataLakeFileClient(self.dsc.url,
self.file_system_name,
file_name,
credential=token_credential)
response = await file_client.create_file()
# Assert
self.assertIsNotNone(response)
@record
def test_create_file_using_oauth_token_credential_async(self):
loop = asyncio.get_event_loop()
loop.run_until_complete(
self._test_create_file_using_oauth_token_credential())
async def _test_create_file_with_existing_name(self):
# Arrange
file_client = await self._create_file_and_return_client()
with self.assertRaises(ResourceExistsError):
# if the file exists then throw error
# if_none_match='*' is to make sure no existing file
await file_client.create_file(
match_condition=MatchConditions.IfMissing)
@record
def test_create_file_with_existing_name_async(self):
loop = asyncio.get_event_loop()
loop.run_until_complete(self._test_create_file_with_existing_name())
async def _test_create_file_with_lease_id(self):
# Arrange
directory_name = self._get_directory_reference()
directory_client = self.dsc.get_directory_client(
self.file_system_name, directory_name)
await directory_client.create_directory()
file_client = directory_client.get_file_client('filename')
# Act
await file_client.create_file()
lease = await file_client.acquire_lease()
create_resp = await file_client.create_file(lease=lease)
# Assert
file_properties = await file_client.get_file_properties()
self.assertIsNotNone(file_properties)
self.assertEqual(file_properties.etag, create_resp.get('etag'))
self.assertEqual(file_properties.last_modified,
create_resp.get('last_modified'))
@record
def test_create_file_with_lease_id_async(self):
loop = asyncio.get_event_loop()
loop.run_until_complete(self._test_create_file_with_lease_id())
async def _test_create_file_under_root_directory(self):
# Arrange
# get a file client to interact with the file under root directory
file_client = self.dsc.get_file_client(self.file_system_name,
"filename")
response = await file_client.create_file()
# Assert
self.assertIsNotNone(response)
@record
def test_create_file_under_root_directory_async(self):
loop = asyncio.get_event_loop()
loop.run_until_complete(self._test_create_file_under_root_directory())
async def _test_append_data(self):
directory_name = self._get_directory_reference()
# Create a directory to put the file under that
directory_client = self.dsc.get_directory_client(
self.file_system_name, directory_name)
await directory_client.create_directory()
file_client = directory_client.get_file_client('filename')
await file_client.create_file()
# Act
response = await file_client.append_data(b'abc', 0, 3)
self.assertIsNotNone(response)
@record
def test_append_data_async(self):
loop = asyncio.get_event_loop()
loop.run_until_complete(self._test_append_data())
async def _test_append_empty_data(self):
file_client = await self._create_file_and_return_client()
# Act
await file_client.flush_data(0)
file_props = await file_client.get_file_properties()
self.assertIsNotNone(file_props['size'], 0)
@record
def test_append_empty_data_async(self):
loop = asyncio.get_event_loop()
loop.run_until_complete(self._test_append_empty_data())
async def _test_flush_data(self):
directory_name = self._get_directory_reference()
# Create a directory to put the file under that
directory_client = self.dsc.get_directory_client(
self.file_system_name, directory_name)
await directory_client.create_directory()
file_client = directory_client.get_file_client('filename')
await file_client.create_file()
# Act
await file_client.append_data(b'abc', 0, 3)
response = await file_client.flush_data(3)
# Assert
prop = await file_client.get_file_properties()
self.assertIsNotNone(response)
self.assertEqual(prop['size'], 3)
@record
def test_flush_data_async(self):
loop = asyncio.get_event_loop()
loop.run_until_complete(self._test_flush_data())
async def _test_flush_data_with_match_condition(self):
directory_name = self._get_directory_reference()
# Create a directory to put the file under that
directory_client = self.dsc.get_directory_client(
self.file_system_name, directory_name)
await directory_client.create_directory()
file_client = directory_client.get_file_client('filename')
resp = await file_client.create_file()
# Act
await file_client.append_data(b'abc', 0, 3)
# flush is successful because it isn't touched
response = await file_client.flush_data(
3,
etag=resp['etag'],
match_condition=MatchConditions.IfNotModified)
await file_client.append_data(b'abc', 3, 3)
with self.assertRaises(ResourceModifiedError):
# flush is unsuccessful because extra data were appended.
await file_client.flush_data(
6,
etag=resp['etag'],
match_condition=MatchConditions.IfNotModified)
@record
def test_flush_data_with_match_condition_async(self):
loop = asyncio.get_event_loop()
loop.run_until_complete(self._test_flush_data_with_match_condition())
async def _test_upload_data(self):
# parallel upload cannot be recorded
if TestMode.need_recording_file(self.test_mode):
return
directory_name = self._get_directory_reference()
# Create a directory to put the file under that
directory_client = self.dsc.get_directory_client(
self.file_system_name, directory_name)
await directory_client.create_directory()
file_client = directory_client.get_file_client('filename')
data = self.get_random_bytes(400 * 1024)
await file_client.upload_data(data, overwrite=True, max_concurrency=5)
downloaded_data = await (await file_client.download_file()).readall()
self.assertEqual(data, downloaded_data)
def test_upload_data_async(self):
loop = asyncio.get_event_loop()
loop.run_until_complete(self._test_upload_data())
async def _test_upload_data_to_existing_file_async(self):
directory_name = self._get_directory_reference()
# Create a directory to put the file under that
directory_client = self.dsc.get_directory_client(
self.file_system_name, directory_name)
await directory_client.create_directory()
# create an existing file
file_client = directory_client.get_file_client('filename')
await file_client.create_file()
await file_client.append_data(b"abc", 0)
await file_client.flush_data(3)
# to override the existing file
data = self.get_random_bytes(100)
with self.assertRaises(HttpResponseError):
await file_client.upload_data(data, max_concurrency=5)
await file_client.upload_data(data, overwrite=True, max_concurrency=5)
downloaded_data = await (await file_client.download_file()).readall()
self.assertEqual(data, downloaded_data)
@record
def test_upload_data_to_existing_file_async(self):
loop = asyncio.get_event_loop()
loop.run_until_complete(
self._test_upload_data_to_existing_file_async())
async def _test_upload_data_to_existing_file_with_content_settings_async(
self):
# etag in async recording file cannot be parsed properly
if TestMode.need_recording_file(self.test_mode):
return
directory_name = self._get_directory_reference()
# Create a directory to put the file under that
directory_client = self.dsc.get_directory_client(
self.file_system_name, directory_name)
await directory_client.create_directory()
# create an existing file
file_client = directory_client.get_file_client('filename')
resp = await file_client.create_file()
etag = resp['etag']
# to override the existing file
data = self.get_random_bytes(100)
content_settings = ContentSettings(content_language='spanish',
content_disposition='inline')
await file_client.upload_data(
data,
max_concurrency=5,
content_settings=content_settings,
etag=etag,
match_condition=MatchConditions.IfNotModified)
downloaded_data = await (await file_client.download_file()).readall()
properties = await file_client.get_file_properties()
self.assertEqual(data, downloaded_data)
self.assertEqual(properties.content_settings.content_language,
content_settings.content_language)
@record
def test_upload_data_to_existing_file_with_content_settings_async(self):
loop = asyncio.get_event_loop()
loop.run_until_complete(
self.
_test_upload_data_to_existing_file_with_content_settings_async())
async def _test_read_file(self):
file_client = await self._create_file_and_return_client()
data = self.get_random_bytes(1024)
# upload data to file
await file_client.append_data(data, 0, len(data))
await file_client.flush_data(len(data))
# doanload the data and make sure it is the same as uploaded data
downloaded_data = await (await file_client.download_file()).readall()
self.assertEqual(data, downloaded_data)
@record
def test_read_file_async(self):
loop = asyncio.get_event_loop()
loop.run_until_complete(self._test_read_file())
async def _test_read_file_with_user_delegation_key(self):
# SAS URL is calculated from storage key, so this test runs live only
if TestMode.need_recording_file(self.test_mode):
return
# Create file
file_client = await self._create_file_and_return_client()
data = self.get_random_bytes(1024)
# Upload data to file
await file_client.append_data(data, 0, len(data))
await file_client.flush_data(len(data))
# Get user delegation key
token_credential = self.generate_async_oauth_token()
service_client = DataLakeServiceClient(self._get_oauth_account_url(),
credential=token_credential)
user_delegation_key = await service_client.get_user_delegation_key(
datetime.utcnow(),
datetime.utcnow() + timedelta(hours=1))
sas_token = generate_file_sas(
file_client.account_name,
file_client.file_system_name,
None,
file_client.path_name,
user_delegation_key,
permission=FileSasPermissions(read=True,
create=True,
write=True,
delete=True),
expiry=datetime.utcnow() + timedelta(hours=1),
)
# doanload the data and make sure it is the same as uploaded data
new_file_client = DataLakeFileClient(self._get_account_url(),
file_client.file_system_name,
file_client.path_name,
credential=sas_token)
downloaded_data = await (await
new_file_client.download_file()).readall()
self.assertEqual(data, downloaded_data)
@record
def test_read_file_with_user_delegation_key_async(self):
loop = asyncio.get_event_loop()
loop.run_until_complete(
self._test_read_file_with_user_delegation_key())
async def _test_read_file_into_file(self):
file_client = await self._create_file_and_return_client()
data = self.get_random_bytes(1024)
# upload data to file
await file_client.append_data(data, 0, len(data))
await file_client.flush_data(len(data))
# doanload the data into a file and make sure it is the same as uploaded data
with open(FILE_PATH, 'wb') as stream:
download = await file_client.download_file(max_concurrency=2)
await download.readinto(stream)
# Assert
with open(FILE_PATH, 'rb') as stream:
actual = stream.read()
self.assertEqual(data, actual)
@record
def test_read_file_into_file_async(self):
loop = asyncio.get_event_loop()
loop.run_until_complete(self._test_read_file_into_file())
async def _test_read_file_to_text(self):
file_client = await self._create_file_and_return_client()
data = self.get_random_text_data(1024)
# upload data to file
await file_client.append_data(data, 0, len(data))
await file_client.flush_data(len(data))
# doanload the text data and make sure it is the same as uploaded data
downloaded_data = await (await file_client.download_file(
max_concurrency=2, encoding="utf-8")).readall()
# Assert
self.assertEqual(data, downloaded_data)
@record
def test_read_file_to_text_async(self):
loop = asyncio.get_event_loop()
loop.run_until_complete(self._test_read_file_to_text())
async def _test_account_sas(self):
# SAS URL is calculated from storage key, so this test runs live only
if TestMode.need_recording_file(self.test_mode):
return
file_name = self._get_file_reference()
# create a file under root directory
await self._create_file_and_return_client(file=file_name)
# generate a token with file level read permission
token = generate_account_sas(
self.dsc.account_name,
self.dsc.credential.account_key,
ResourceTypes(file_system=True, object=True),
AccountSasPermissions(read=True),
datetime.utcnow() + timedelta(hours=1),
)
# read the created file which is under root directory
file_client = DataLakeFileClient(self.dsc.url,
self.file_system_name,
file_name,
credential=token)
properties = await file_client.get_file_properties()
# make sure we can read the file properties
self.assertIsNotNone(properties)
# try to write to the created file with the token
with self.assertRaises(HttpResponseError):
await file_client.append_data(b"abcd", 0, 4)
@record
def test_account_sas_async(self):
loop = asyncio.get_event_loop()
loop.run_until_complete(self._test_account_sas())
async def _test_file_sas_only_applies_to_file_level(self):
# SAS URL is calculated from storage key, so this test runs live only
if TestMode.need_recording_file(self.test_mode):
return
file_name = self._get_file_reference()
directory_name = self._get_directory_reference()
await self._create_file_and_return_client(directory=directory_name,
file=file_name)
# generate a token with file level read and write permissions
token = generate_file_sas(
self.dsc.account_name,
self.file_system_name,
directory_name,
file_name,
self.dsc.credential.account_key,
permission=FileSasPermissions(read=True, write=True),
expiry=datetime.utcnow() + timedelta(hours=1),
)
# read the created file which is under root directory
file_client = DataLakeFileClient(self.dsc.url,
self.file_system_name,
directory_name + '/' + file_name,
credential=token)
properties = await file_client.get_file_properties()
# make sure we can read the file properties
self.assertIsNotNone(properties)
# try to write to the created file with the token
response = await file_client.append_data(b"abcd",
0,
4,
validate_content=True)
self.assertIsNotNone(response)
# the token is for file level, so users are not supposed to have access to file system level operations
file_system_client = FileSystemClient(self.dsc.url,
self.file_system_name,
credential=token)
with self.assertRaises(ClientAuthenticationError):
await file_system_client.get_file_system_properties()
# the token is for file level, so users are not supposed to have access to directory level operations
directory_client = DataLakeDirectoryClient(self.dsc.url,
self.file_system_name,
directory_name,
credential=token)
with self.assertRaises(ClientAuthenticationError):
await directory_client.get_directory_properties()
@record
def test_file_sas_only_applies_to_file_level_async(self):
loop = asyncio.get_event_loop()
loop.run_until_complete(
self._test_file_sas_only_applies_to_file_level())
async def _test_delete_file(self):
# Arrange
file_client = await self._create_file_and_return_client()
await file_client.delete_file()
with self.assertRaises(ResourceNotFoundError):
await file_client.get_file_properties()
@record
def test_delete_file_async(self):
loop = asyncio.get_event_loop()
loop.run_until_complete(self._test_delete_file())
async def _test_delete_file_with_if_unmodified_since(self):
# Arrange
file_client = await self._create_file_and_return_client()
prop = await file_client.get_file_properties()
await file_client.delete_file(if_unmodified_since=prop['last_modified']
)
# Make sure the file was deleted
with self.assertRaises(ResourceNotFoundError):
await file_client.get_file_properties()
@record
def test_delete_file_with_if_unmodified_since_async(self):
loop = asyncio.get_event_loop()
loop.run_until_complete(
self._test_delete_file_with_if_unmodified_since())
async def _test_set_access_control(self):
file_client = await self._create_file_and_return_client()
response = await file_client.set_access_control(permissions='0777')
# Assert
self.assertIsNotNone(response)
@record
def test_set_access_control_async(self):
loop = asyncio.get_event_loop()
loop.run_until_complete(self._test_set_access_control())
async def _test_set_access_control_with_match_conditions(self):
file_client = await self._create_file_and_return_client()
with self.assertRaises(ResourceModifiedError):
await file_client.set_access_control(
permissions='0777', match_condition=MatchConditions.IfMissing)
@record
def test_set_access_control_with_match_conditions_async(self):
loop = asyncio.get_event_loop()
loop.run_until_complete(
self._test_set_access_control_with_match_conditions())
async def _test_get_access_control(self):
file_client = await self._create_file_and_return_client()
await file_client.set_access_control(permissions='0777')
# Act
response = await file_client.get_access_control()
# Assert
self.assertIsNotNone(response)
@record
def test_get_access_control_async(self):
loop = asyncio.get_event_loop()
loop.run_until_complete(self._test_get_access_control())
async def _test_get_access_control_with_if_modified_since(self):
file_client = await self._create_file_and_return_client()
await file_client.set_access_control(permissions='0777')
prop = await file_client.get_file_properties()
# Act
response = await file_client.get_access_control(
if_modified_since=prop['last_modified'] - timedelta(minutes=15))
# Assert
self.assertIsNotNone(response)
@record
def test_get_access_control_with_if_modified_since_async(self):
loop = asyncio.get_event_loop()
loop.run_until_complete(
self._test_get_access_control_with_if_modified_since())
async def _test_get_properties(self):
# Arrange
directory_client = await self._create_directory_and_return_client()
metadata = {'hello': 'world', 'number': '42'}
content_settings = ContentSettings(content_language='spanish',
content_disposition='inline')
file_client = await directory_client.create_file(
"newfile", metadata=metadata, content_settings=content_settings)
await file_client.append_data(b"abc", 0, 3)
await file_client.flush_data(3)
properties = await file_client.get_file_properties()
# Assert
self.assertTrue(properties)
self.assertEqual(properties.size, 3)
self.assertEqual(properties.metadata['hello'], metadata['hello'])
self.assertEqual(properties.content_settings.content_language,
content_settings.content_language)
@record
def test_get_properties_async(self):
loop = asyncio.get_event_loop()
loop.run_until_complete(self._test_get_properties())
async def _test_rename_file_with_non_used_name(self):
file_client = await self._create_file_and_return_client()
data_bytes = b"abc"
await file_client.append_data(data_bytes, 0, 3)
await file_client.flush_data(3)
new_client = await file_client.rename_file(
file_client.file_system_name + '/' + 'newname')
data = await (await new_client.download_file()).readall()
self.assertEqual(data, data_bytes)
self.assertEqual(new_client.path_name, "newname")
@record
def test_rename_file_with_non_used_name_async(self):
loop = asyncio.get_event_loop()
loop.run_until_complete(self._test_rename_file_with_non_used_name())
async def _test_rename_file_to_existing_file(self):
# create the existing file
existing_file_client = await self._create_file_and_return_client(
file="existingfile")
await existing_file_client.append_data(b"a", 0, 1)
await existing_file_client.flush_data(1)
old_url = existing_file_client.url
# prepare to rename the file to the existing file
file_client = await self._create_file_and_return_client()
data_bytes = b"abc"
await file_client.append_data(data_bytes, 0, 3)
await file_client.flush_data(3)
new_client = await file_client.rename_file(
file_client.file_system_name + '/' +
existing_file_client.path_name)
new_url = file_client.url
data = await (await new_client.download_file()).readall()
# the existing file was overridden
self.assertEqual(data, data_bytes)
@record
def test_rename_file_to_existing_file_async(self):
loop = asyncio.get_event_loop()
loop.run_until_complete(self._test_rename_file_to_existing_file())
async def _test_rename_file_will_not_change_existing_directory(self):
if TestMode.need_recording_file(self.test_mode):
return
# create none empty directory(with 2 files)
dir1 = await self._create_directory_and_return_client(directory="dir1")
f1 = await dir1.create_file("file1")
await f1.append_data(b"file1", 0, 5)
await f1.flush_data(5)
f2 = await dir1.create_file("file2")
await f2.append_data(b"file2", 0, 5)
await f2.flush_data(5)
# create another none empty directory(with 2 files)
dir2 = await self._create_directory_and_return_client(directory="dir2")
f3 = await dir2.create_file("file3")
await f3.append_data(b"file3", 0, 5)
await f3.flush_data(5)
f4 = await dir2.create_file("file4")
await f4.append_data(b"file4", 0, 5)
await f4.flush_data(5)
new_client = await f3.rename_file(f1.file_system_name + '/' +
f1.path_name)
self.assertEqual(await (await new_client.download_file()).readall(),
b"file3")
# make sure the data in file2 and file4 weren't touched
f2_data = await (await f2.download_file()).readall()
self.assertEqual(f2_data, b"file2")
f4_data = await (await f4.download_file()).readall()
self.assertEqual(f4_data, b"file4")
with self.assertRaises(HttpResponseError):
await (await f3.download_file()).readall()
@record
def test_rename_file_will_not_change_existing_directory_async(self):
loop = asyncio.get_event_loop()
loop.run_until_complete(
self._test_rename_file_will_not_change_existing_directory())
def _setUp(self, account_name, account_key):
url = self._get_account_url(account_name)
self.dsc = DataLakeServiceClient(url, credential=account_key, logging_enable=True)
self.config = self.dsc._config
class FileSystemTest(StorageTestCase):
def _setUp(self, account_name, account_key):
url = self._get_account_url(account_name)
self.dsc = DataLakeServiceClient(url, credential=account_key,
transport=AiohttpTestTransport(), logging_enable=True)
self.config = self.dsc._config
self.test_file_systems = []
def tearDown(self):
if not self.is_playback():
loop = asyncio.get_event_loop()
try:
for file_system in self.test_file_systems:
loop.run_until_complete(self.dsc.delete_file_system(file_system))
loop.run_until_complete(self.fsc.__aexit__())
except:
pass
return super(FileSystemTest, self).tearDown()
# --Helpers-----------------------------------------------------------------
def _get_file_system_reference(self, prefix=TEST_FILE_SYSTEM_PREFIX):
file_system_name = self.get_resource_name(prefix)
self.test_file_systems.append(file_system_name)
return file_system_name
async def _create_file_system(self, file_system_prefix=TEST_FILE_SYSTEM_PREFIX):
return await self.dsc.create_file_system(self._get_file_system_reference(prefix=file_system_prefix))
# --Helpers-----------------------------------------------------------------
@DataLakePreparer()
async def test_create_file_system_async(self, datalake_storage_account_name, datalake_storage_account_key):
self._setUp(datalake_storage_account_name, datalake_storage_account_key)
# Arrange
file_system_name = self._get_file_system_reference()
# Act
file_system_client = self.dsc.get_file_system_client(file_system_name)
created = await file_system_client.create_file_system()
# Assert
self.assertTrue(created)
@DataLakePreparer()
async def test_file_system_exists(self, datalake_storage_account_name, datalake_storage_account_key):
self._setUp(datalake_storage_account_name, datalake_storage_account_key)
# Arrange
file_system_name = self._get_file_system_reference()
# Act
file_system_client1 = self.dsc.get_file_system_client(file_system_name)
file_system_client2 = self.dsc.get_file_system_client("nonexistentfs")
await file_system_client1.create_file_system()
self.assertTrue(await file_system_client1.exists())
self.assertFalse(await file_system_client2.exists())
@DataLakePreparer()
async def test_create_file_system_with_metadata_async(
self, datalake_storage_account_name, datalake_storage_account_key):
self._setUp(datalake_storage_account_name, datalake_storage_account_key)
# Arrange
metadata = {'hello': 'world', 'number': '42'}
file_system_name = self._get_file_system_reference()
# Act
file_system_client = self.dsc.get_file_system_client(file_system_name)
created = await file_system_client.create_file_system(metadata=metadata)
# Assert
properties = await file_system_client.get_file_system_properties()
self.assertTrue(created)
self.assertDictEqual(properties.metadata, metadata)
@DataLakePreparer()
async def test_list_file_systems_async(self, datalake_storage_account_name, datalake_storage_account_key):
self._setUp(datalake_storage_account_name, datalake_storage_account_key)
# Arrange
file_system_name = self._get_file_system_reference()
file_system = await self.dsc.create_file_system(file_system_name)
# Act
file_systems = []
async for filesystem in self.dsc.list_file_systems():
file_systems.append(filesystem)
# Assert
self.assertIsNotNone(file_systems)
self.assertGreaterEqual(len(file_systems), 1)
self.assertIsNotNone(file_systems[0])
self.assertNamedItemInContainer(file_systems, file_system.file_system_name)
self.assertIsNotNone(file_systems[0].has_immutability_policy)
self.assertIsNotNone(file_systems[0].has_legal_hold)
@DataLakePreparer()
async def test_delete_file_system_with_existing_file_system_async(
self, datalake_storage_account_name, datalake_storage_account_key):
self._setUp(datalake_storage_account_name, datalake_storage_account_key)
# Arrange
file_system = await self._create_file_system()
# Act
deleted = await file_system.delete_file_system()
# Assert
self.assertIsNone(deleted)
@DataLakePreparer()
async def test_rename_file_system(self, datalake_storage_account_name, datalake_storage_account_key):
if not self.is_playback():
return
self._setUp(datalake_storage_account_name, datalake_storage_account_key)
old_name1 = self._get_file_system_reference(prefix="oldcontainer1")
old_name2 = self._get_file_system_reference(prefix="oldcontainer2")
new_name = self._get_file_system_reference(prefix="newcontainer")
filesystem1 = await self.dsc.create_file_system(old_name1)
await self.dsc.create_file_system(old_name2)
new_filesystem = await self.dsc._rename_file_system(name=old_name1, new_name=new_name)
with self.assertRaises(HttpResponseError):
await self.dsc._rename_file_system(name=old_name2, new_name=new_name)
with self.assertRaises(HttpResponseError):
await filesystem1.get_file_system_properties()
with self.assertRaises(HttpResponseError):
await self.dsc._rename_file_system(name="badfilesystem", new_name="filesystem")
props = await new_filesystem.get_file_system_properties()
self.assertEqual(new_name, props.name)
@DataLakePreparer()
async def test_rename_file_system_with_file_system_client(
self, datalake_storage_account_name, datalake_storage_account_key):
pytest.skip("Feature not yet enabled. Make sure to record this test once enabled.")
self._setUp(datalake_storage_account_name, datalake_storage_account_key)
old_name1 = self._get_file_system_reference(prefix="oldcontainer1")
old_name2 = self._get_file_system_reference(prefix="oldcontainer2")
new_name = self._get_file_system_reference(prefix="newcontainer")
bad_name = self._get_file_system_reference(prefix="badcontainer")
filesystem1 = await self.dsc.create_file_system(old_name1)
file_system2 = await self.dsc.create_file_system(old_name2)
bad_file_system = self.dsc.get_file_system_client(bad_name)
new_filesystem = await filesystem1._rename_file_system(new_name=new_name)
with self.assertRaises(HttpResponseError):
await file_system2._rename_file_system(new_name=new_name)
with self.assertRaises(HttpResponseError):
await filesystem1.get_file_system_properties()
with self.assertRaises(HttpResponseError):
await bad_file_system._rename_file_system(new_name="filesystem")
new_file_system_props = await new_filesystem.get_file_system_properties()
self.assertEqual(new_name, new_file_system_props.name)
@DataLakePreparer()
async def test_rename_file_system_with_source_lease(
self, datalake_storage_account_name, datalake_storage_account_key):
if not self.is_playback():
return
self._setUp(datalake_storage_account_name, datalake_storage_account_key)
old_name = self._get_file_system_reference(prefix="old")
new_name = self._get_file_system_reference(prefix="new")
filesystem = await self.dsc.create_file_system(old_name)
filesystem_lease_id = await filesystem.acquire_lease()
with self.assertRaises(HttpResponseError):
await self.dsc._rename_file_system(name=old_name, new_name=new_name)
with self.assertRaises(HttpResponseError):
await self.dsc._rename_file_system(name=old_name, new_name=new_name, lease="bad_id")
new_filesystem = await self.dsc._rename_file_system(name=old_name, new_name=new_name, lease=filesystem_lease_id)
props = await new_filesystem.get_file_system_properties()
self.assertEqual(new_name, props.name)
@DataLakePreparer()
async def test_undelete_file_system(self, datalake_storage_account_name, datalake_storage_account_key):
# Needs soft delete enabled account.
if not self.is_playback():
return
self._setUp(datalake_storage_account_name, datalake_storage_account_key)
name = self._get_file_system_reference(prefix="filesystem")
filesystem_client = await self.dsc.create_file_system(name)
await filesystem_client.delete_file_system()
# to make sure the filesystem deleted
with self.assertRaises(ResourceNotFoundError):
await filesystem_client.get_file_system_properties()
filesystem_list = []
async for fs in self.dsc.list_file_systems(include_deleted=True):
filesystem_list.append(fs)
self.assertTrue(len(filesystem_list) >= 1)
restored_version = 0
for filesystem in filesystem_list:
# find the deleted filesystem and restore it
if filesystem.deleted and filesystem.name == filesystem_client.file_system_name:
restored_fs_client = await self.dsc.undelete_file_system(filesystem.name,
filesystem.deleted_version,
new_name="restored" +
name + str(restored_version))
restored_version += 1
# to make sure the deleted filesystem is restored
props = await restored_fs_client.get_file_system_properties()
self.assertIsNotNone(props)
@DataLakePreparer()
async def test_restore_to_existing_file_system(self, datalake_storage_account_name, datalake_storage_account_key):
# Needs soft delete enabled account.
if not self.is_playback():
return
self._setUp(datalake_storage_account_name, datalake_storage_account_key)
# get an existing filesystem
existing_name = self._get_file_system_reference(prefix="existing")
name = self._get_file_system_reference(prefix="filesystem")
existing_filesystem_client = await self.dsc.create_file_system(existing_name)
filesystem_client = await self.dsc.create_file_system(name)
# Act
await filesystem_client.delete_file_system()
# to make sure the filesystem deleted
with self.assertRaises(ResourceNotFoundError):
await filesystem_client.get_file_system_properties()
filesystem_list = []
async for fs in self.dsc.list_file_systems(include_deleted=True):
filesystem_list.append(fs)
self.assertTrue(len(filesystem_list) >= 1)
for filesystem in filesystem_list:
# find the deleted filesystem and restore it
if filesystem.deleted and filesystem.name == filesystem_client.file_system_name:
with self.assertRaises(HttpResponseError):
await self.dsc.undelete_file_system(filesystem.name, filesystem.deleted_version,
new_name=existing_filesystem_client.file_system_name)
@DataLakePreparer()
async def test_restore_file_system_with_sas(self, datalake_storage_account_name, datalake_storage_account_key):
pytest.skip(
"We are generating a SAS token therefore play only live but we also need a soft delete enabled account.")
self._setUp(datalake_storage_account_name, datalake_storage_account_key)
token = generate_account_sas(
self.dsc.account_name,
self.dsc.credential.account_key,
ResourceTypes(service=True, file_system=True),
AccountSasPermissions(read=True, write=True, list=True, delete=True),
datetime.utcnow() + timedelta(hours=1),
)
dsc = DataLakeServiceClient(self.dsc.url, token)
name = self._get_file_system_reference(prefix="filesystem")
filesystem_client = await dsc.create_file_system(name)
await filesystem_client.delete_file_system()
# to make sure the filesystem is deleted
with self.assertRaises(ResourceNotFoundError):
await filesystem_client.get_file_system_properties()
filesystem_list = []
async for fs in self.dsc.list_file_systems(include_deleted=True):
filesystem_list.append(fs)
self.assertTrue(len(filesystem_list) >= 1)
restored_version = 0
for filesystem in filesystem_list:
# find the deleted filesystem and restore it
if filesystem.deleted and filesystem.name == filesystem_client.file_system_name:
restored_fs_client = await dsc.undelete_file_system(filesystem.name, filesystem.deleted_version,
new_name="restored" + name + str(restored_version))
restored_version += 1
# to make sure the deleted filesystem is restored
props = await restored_fs_client.get_file_system_properties()
self.assertIsNotNone(props)
@DataLakePreparer()
async def test_delete_none_existing_file_system_async(
self, datalake_storage_account_name, datalake_storage_account_key):
self._setUp(datalake_storage_account_name, datalake_storage_account_key)
fake_file_system_client = self.dsc.get_file_system_client("fakeclient")
# Act
with self.assertRaises(ResourceNotFoundError):
await fake_file_system_client.delete_file_system(match_condition=MatchConditions.IfMissing)
@DataLakePreparer()
async def test_list_file_systems_with_include_metadata_async(
self, datalake_storage_account_name, datalake_storage_account_key):
self._setUp(datalake_storage_account_name, datalake_storage_account_key)
# Arrange
file_system = await self._create_file_system()
metadata = {'hello': 'world', 'number': '42'}
await file_system.set_file_system_metadata(metadata)
# Act
file_systems = []
async for fs in self.dsc.list_file_systems(name_starts_with=file_system.file_system_name,
include_metadata=True):
file_systems.append(fs)
# Assert
self.assertIsNotNone(file_systems)
self.assertGreaterEqual(len(file_systems), 1)
self.assertIsNotNone(file_systems[0])
self.assertNamedItemInContainer(file_systems, file_system.file_system_name)
self.assertDictEqual(file_systems[0].metadata, metadata)
@DataLakePreparer()
async def test_set_file_system_acl_async(self, datalake_storage_account_name, datalake_storage_account_key):
self._setUp(datalake_storage_account_name, datalake_storage_account_key)
# Act
file_system = await self._create_file_system()
access_policy = AccessPolicy(permission=FileSystemSasPermissions(read=True),
expiry=datetime.utcnow() + timedelta(hours=1),
start=datetime.utcnow())
signed_identifier1 = {'testid': access_policy}
response = await file_system.set_file_system_access_policy(
signed_identifier1, public_access=PublicAccess.FileSystem)
self.assertIsNotNone(response.get('etag'))
self.assertIsNotNone(response.get('last_modified'))
acl1 = await file_system.get_file_system_access_policy()
self.assertIsNotNone(acl1['public_access'])
self.assertEqual(len(acl1['signed_identifiers']), 1)
# If set signed identifier without specifying the access policy then it will be default to None
signed_identifier2 = {'testid': access_policy, 'test2': access_policy}
await file_system.set_file_system_access_policy(signed_identifier2)
acl2 = await file_system.get_file_system_access_policy()
self.assertIsNone(acl2['public_access'])
self.assertEqual(len(acl2['signed_identifiers']), 2)
@DataLakePreparer()
async def test_list_file_systems_by_page_async(self, datalake_storage_account_name, datalake_storage_account_key):
self._setUp(datalake_storage_account_name, datalake_storage_account_key)
# Arrange
for i in range(0, 6):
await self._create_file_system(file_system_prefix="filesystem{}".format(i))
# Act
file_systems = []
async for fs in await self.dsc.list_file_systems(
results_per_page=3,
name_starts_with="file",
include_metadata=True).by_page().__anext__():
file_systems.append(fs)
# Assert
self.assertIsNotNone(file_systems)
self.assertGreaterEqual(len(file_systems), 3)
@DataLakePreparer()
async def test_list_file_systems_with_public_access_async(
self, datalake_storage_account_name, datalake_storage_account_key):
self._setUp(datalake_storage_account_name, datalake_storage_account_key)
# Arrange
file_system_name = self._get_file_system_reference()
file_system = self.dsc.get_file_system_client(file_system_name)
await file_system.create_file_system(public_access="blob")
metadata = {'hello': 'world', 'number': '42'}
await file_system.set_file_system_metadata(metadata)
# Act
file_systems = []
async for fs in self.dsc.list_file_systems(name_starts_with=file_system.file_system_name,
include_metadata=True):
file_systems.append(fs)
# Assert
self.assertIsNotNone(file_systems)
self.assertGreaterEqual(len(file_systems), 1)
self.assertIsNotNone(file_systems[0])
self.assertNamedItemInContainer(file_systems, file_system.file_system_name)
self.assertDictEqual(file_systems[0].metadata, metadata)
self.assertTrue(file_systems[0].public_access is PublicAccess.File)
@DataLakePreparer()
async def test_get_file_system_properties_async(self, datalake_storage_account_name, datalake_storage_account_key):
self._setUp(datalake_storage_account_name, datalake_storage_account_key)
# Arrange
metadata = {'hello': 'world', 'number': '42'}
file_system = await self._create_file_system()
await file_system.set_file_system_metadata(metadata)
# Act
props = await file_system.get_file_system_properties()
# Assert
self.assertIsNotNone(props)
self.assertDictEqual(props.metadata, metadata)
self.assertIsNotNone(props.has_immutability_policy)
self.assertIsNotNone(props.has_legal_hold)
@DataLakePreparer()
async def test_service_client_session_closes_after_filesystem_creation(
self, datalake_storage_account_name, datalake_storage_account_key):
self._setUp(datalake_storage_account_name, datalake_storage_account_key)
# Arrange
dsc2 = DataLakeServiceClient(self.dsc.url, credential=datalake_storage_account_key)
async with DataLakeServiceClient(
self.dsc.url, credential=datalake_storage_account_key) as ds_client:
fs1 = await ds_client.create_file_system(self._get_file_system_reference(prefix="fs1"))
await fs1.delete_file_system()
await dsc2.create_file_system(self._get_file_system_reference(prefix="fs2"))
await dsc2.close()
@DataLakePreparer()
async def test_list_paths_async(self, datalake_storage_account_name, datalake_storage_account_key):
self._setUp(datalake_storage_account_name, datalake_storage_account_key)
# Arrange
file_system = await self._create_file_system()
for i in range(0, 6):
await file_system.create_directory("dir1{}".format(i))
paths = []
async for path in file_system.get_paths(upn=True):
paths.append(path)
self.assertEqual(len(paths), 6)
@DataLakePreparer()
async def test_list_paths_which_are_all_files_async(
self, datalake_storage_account_name, datalake_storage_account_key):
self._setUp(datalake_storage_account_name, datalake_storage_account_key)
# Arrange
file_system = await self._create_file_system()
for i in range(0, 6):
await file_system.create_file("file{}".format(i))
paths = []
async for path in file_system.get_paths(upn=True):
paths.append(path)
self.assertEqual(len(paths), 6)
@DataLakePreparer()
async def test_list_paths_with_max_per_page_async(
self, datalake_storage_account_name, datalake_storage_account_key):
self._setUp(datalake_storage_account_name, datalake_storage_account_key)
# Arrange
file_system = await self._create_file_system()
for i in range(0, 6):
await file_system.create_directory("dir1{}".format(i))
generator1 = file_system.get_paths(max_results=2, upn=True).by_page()
paths1 = []
async for path in await generator1.__anext__():
paths1.append(path)
generator2 = file_system.get_paths(max_results=4, upn=True) \
.by_page(continuation_token=generator1.continuation_token)
paths2 = []
async for path in await generator2.__anext__():
paths2.append(path)
self.assertEqual(len(paths1), 2)
self.assertEqual(len(paths2), 4)
@DataLakePreparer()
async def test_list_paths_under_specific_path_async(
self, datalake_storage_account_name, datalake_storage_account_key):
self._setUp(datalake_storage_account_name, datalake_storage_account_key)
# Arrange
file_system = await self._create_file_system()
for i in range(0, 6):
dir = await file_system.create_directory("dir1{}".format(i))
# create a subdirectory under the current directory
subdir = await dir.create_sub_directory("subdir")
await subdir.create_sub_directory("subsub")
# create a file under the current directory
file_client = await subdir.create_file("file")
await file_client.append_data(b"abced", 0, 5)
await file_client.flush_data(5)
generator1 = file_system.get_paths(path="dir10/subdir", max_results=2, upn=True).by_page()
paths = []
async for path in await generator1.__anext__():
paths.append(path)
self.assertEqual(len(paths), 2)
self.assertEqual(paths[0].content_length, 5)
@DataLakePreparer()
async def test_list_paths_recursively_async(self, datalake_storage_account_name, datalake_storage_account_key):
self._setUp(datalake_storage_account_name, datalake_storage_account_key)
# Arrange
file_system = await self._create_file_system()
for i in range(0, 6):
dir = await file_system.create_directory("dir1{}".format(i))
# create a subdirectory under the current directory
subdir = await dir.create_sub_directory("subdir")
await subdir.create_sub_directory("subsub")
# create a file under the current directory
await subdir.create_file("file")
paths = []
async for path in file_system.get_paths(recursive=True, upn=True):
paths.append(path)
# there are 24 subpaths in total
self.assertEqual(len(paths), 24)
@DataLakePreparer()
async def test_list_paths_pages_correctly(self, datalake_storage_account_name, datalake_storage_account_key):
self._setUp(datalake_storage_account_name, datalake_storage_account_key)
# Arrange
file_system = await self._create_file_system(file_system_prefix="filesystem1")
for i in range(0, 6):
await file_system.create_directory("dir1{}".format(i))
for i in range(0, 6):
await file_system.create_file("file{}".format(i))
generator = file_system.get_paths(max_results=6, upn=True).by_page()
paths1 = []
async for path in await generator.__anext__():
paths1.append(path)
paths2 = []
async for path in await generator.__anext__():
paths2.append(path)
with self.assertRaises(StopAsyncIteration):
paths3 = []
async for path in await generator.__anext__():
paths3.append(path)
self.assertEqual(len(paths1), 6)
self.assertEqual(len(paths2), 6)
@DataLakePreparer()
async def test_get_deleted_paths(self, datalake_storage_account_name, datalake_storage_account_key):
if not self.is_playback():
return
self._setUp(datalake_storage_account_name, datalake_storage_account_key)
# Arrange
file_system = await self._create_file_system()
file0 = await file_system.create_file("file0")
file1 = await file_system.create_file("file1")
dir1 = await file_system.create_directory("dir1")
dir2 = await file_system.create_directory("dir2")
dir3 = await file_system.create_directory("dir3")
file_in_dir3 = await dir3.create_file("file_in_dir3")
file_in_subdir = await dir3.create_file("subdir/file_in_subdir")
await file0.delete_file()
await file1.delete_file()
await dir1.delete_directory()
await dir2.delete_directory()
await file_in_dir3.delete_file()
await file_in_subdir.delete_file()
deleted_paths = []
async for path in file_system.list_deleted_paths():
deleted_paths.append(path)
dir3_paths = []
async for path in file_system.list_deleted_paths(path_prefix="dir3/"):
dir3_paths.append(path)
# Assert
self.assertEqual(len(deleted_paths), 6)
self.assertEqual(len(dir3_paths), 2)
self.assertIsNotNone(dir3_paths[0].deletion_id)
self.assertIsNotNone(dir3_paths[1].deletion_id)
self.assertEqual(dir3_paths[0].name, 'dir3/file_in_dir3')
self.assertEqual(dir3_paths[1].name, 'dir3/subdir/file_in_subdir')
paths_generator1 = file_system.list_deleted_paths(results_per_page=2).by_page()
paths1 = []
async for path in await paths_generator1.__anext__():
paths1.append(path)
paths_generator2 = file_system.list_deleted_paths(results_per_page=4) \
.by_page(continuation_token=paths_generator1.continuation_token)
paths2 = []
async for path in await paths_generator2.__anext__():
paths2.append(path)
# Assert
self.assertEqual(len(paths1), 2)
self.assertEqual(len(paths2), 4)
@DataLakePreparer()
async def test_create_directory_from_file_system_client_async(
self, datalake_storage_account_name, datalake_storage_account_key):
self._setUp(datalake_storage_account_name, datalake_storage_account_key)
# Arrange
file_system = await self._create_file_system()
await file_system.create_directory("dir1/dir2")
paths = []
async for path in file_system.get_paths(recursive=False, upn=True):
paths.append(path)
self.assertEqual(len(paths), 1)
self.assertEqual(paths[0].name, "dir1")
@DataLakePreparer()
async def test_create_file_from_file_system_client_async(
self, datalake_storage_account_name, datalake_storage_account_key):
self._setUp(datalake_storage_account_name, datalake_storage_account_key)
# Arrange
file_system = await self._create_file_system()
await file_system.create_file("dir1/dir2/file")
paths = []
async for path in file_system.get_paths(recursive=True, upn=True):
paths.append(path)
self.assertEqual(len(paths), 3)
self.assertEqual(paths[0].name, "dir1")
self.assertEqual(paths[2].is_directory, False)
@DataLakePreparer()
async def test_get_root_directory_client_async(self, datalake_storage_account_name, datalake_storage_account_key):
self._setUp(datalake_storage_account_name, datalake_storage_account_key)
file_system = await self._create_file_system()
directory_client = file_system._get_root_directory_client()
acl = 'user::rwx,group::r-x,other::rwx'
await directory_client.set_access_control(acl=acl)
access_control = await directory_client.get_access_control()
self.assertEqual(acl, access_control['acl'])
@pytest.mark.live_test_only
@DataLakePreparer()
async def test_get_access_control_using_delegation_sas_async(
self, datalake_storage_account_name, datalake_storage_account_key):
self._setUp(datalake_storage_account_name, datalake_storage_account_key)
url = self._get_account_url(datalake_storage_account_name)
token_credential = self.generate_oauth_token()
dsc = DataLakeServiceClient(url, token_credential, logging_enable=True)
file_system_name = self._get_file_system_reference()
directory_client_name = '/'
(await dsc.create_file_system(file_system_name)).get_directory_client(directory_client_name)
directory_client = self.dsc.get_directory_client(file_system_name, directory_client_name)
random_guid = uuid.uuid4()
await directory_client.set_access_control(owner=random_guid,
permissions='0777')
acl = await directory_client.get_access_control()
delegation_key = await dsc.get_user_delegation_key(datetime.utcnow(),
datetime.utcnow() + timedelta(hours=1))
token = generate_file_system_sas(
dsc.account_name,
file_system_name,
delegation_key,
permission=FileSystemSasPermissions(
read=True, execute=True, manage_access_control=True, manage_ownership=True),
expiry=datetime.utcnow() + timedelta(hours=1),
agent_object_id=random_guid
)
sas_directory_client = DataLakeDirectoryClient(self.dsc.url, file_system_name, directory_client_name,
credential=token, logging_enable=True)
access_control = await sas_directory_client.get_access_control()
self.assertIsNotNone(access_control)
@pytest.mark.live_test_only
@DataLakePreparer()
async def test_list_paths_using_file_sys_delegation_sas_async(
self, datalake_storage_account_name, datalake_storage_account_key):
self._setUp(datalake_storage_account_name, datalake_storage_account_key)
url = self._get_account_url(datalake_storage_account_name)
token_credential = self.generate_oauth_token()
dsc = DataLakeServiceClient(url, token_credential)
file_system_name = self._get_file_system_reference()
directory_client_name = '/'
directory_client = (await dsc.create_file_system(file_system_name)).get_directory_client(directory_client_name)
random_guid = uuid.uuid4()
await directory_client.set_access_control(owner=random_guid, permissions='0777')
delegation_key = await dsc.get_user_delegation_key(datetime.utcnow(),
datetime.utcnow() + timedelta(hours=1))
token = generate_file_system_sas(
dsc.account_name,
file_system_name,
delegation_key,
permission=DirectorySasPermissions(list=True),
expiry=datetime.utcnow() + timedelta(hours=1),
agent_object_id=random_guid
)
sas_directory_client = FileSystemClient(self.dsc.url, file_system_name,
credential=token)
paths = list()
async for path in sas_directory_client.get_paths():
paths.append(path)
self.assertEqual(0, 0)
@DataLakePreparer()
async def test_file_system_sessions_closes_properly_async(
self, datalake_storage_account_name, datalake_storage_account_key):
self._setUp(datalake_storage_account_name, datalake_storage_account_key)
# Arrange
file_system_client = await self._create_file_system("fs")
async with file_system_client as fs_client:
async with fs_client.get_file_client("file1.txt") as f_client:
await f_client.create_file()
async with fs_client.get_file_client("file2.txt") as f_client:
await f_client.create_file()
async with fs_client.get_directory_client("file1") as f_client:
await f_client.create_directory()
async with fs_client.get_directory_client("file2") as f_client:
await f_client.create_directory()
@DataLakePreparer()
async def test_undelete_dir_with_version_id(self, datalake_storage_account_name, datalake_storage_account_key):
self._setUp(datalake_storage_account_name, datalake_storage_account_key)
if not self.is_playback():
return
file_system_client = await self._create_file_system("fs")
dir_path = 'dir10'
dir_client = await file_system_client.create_directory(dir_path)
resp = await dir_client.delete_directory()
with self.assertRaises(HttpResponseError):
await file_system_client.get_file_client(dir_path).get_file_properties()
restored_dir_client = await file_system_client._undelete_path(dir_path, resp['deletion_id'])
resp = await restored_dir_client.get_directory_properties()
self.assertIsNotNone(resp)
@DataLakePreparer()
async def test_undelete_file_with_version_id(self, datalake_storage_account_name, datalake_storage_account_key):
self._setUp(datalake_storage_account_name, datalake_storage_account_key)
if not self.is_playback():
return
file_system_client = await self._create_file_system("fs")
file_path = 'dir10/fileŇ'
dir_client = await file_system_client.create_file(file_path)
resp = await dir_client.delete_file()
with self.assertRaises(HttpResponseError):
await file_system_client.get_file_client(file_path).get_file_properties()
restored_file_client = await file_system_client._undelete_path(file_path, resp['deletion_id'])
resp = await restored_file_client.get_file_properties()
self.assertIsNotNone(resp)
class FileTest(StorageTestCase):
async def _setUp(self, account_name, account_key):
url = self.account_url(account_name, 'dfs')
self.dsc = DataLakeServiceClient(url, credential=account_key)
self.config = self.dsc._config
self.file_system_name = self.get_resource_name('filesystem')
if not self.is_playback():
file_system = self.dsc.get_file_system_client(self.file_system_name)
try:
await file_system.create_file_system(timeout=5)
except ResourceExistsError:
pass
def tearDown(self):
if not self.is_playback():
try:
loop = asyncio.get_event_loop()
loop.run_until_complete(self.dsc.delete_file_system(self.file_system_name))
loop.run_until_complete(self.dsc.__aexit__())
except:
pass
return super(FileTest, self).tearDown()
# --Helpers-----------------------------------------------------------------
def _get_directory_reference(self, prefix=TEST_DIRECTORY_PREFIX):
directory_name = self.get_resource_name(prefix)
return directory_name
def _get_file_reference(self, prefix=TEST_FILE_PREFIX):
file_name = self.get_resource_name(prefix)
return file_name
async def _create_file_system(self):
return await self.dsc.create_file_system(self._get_file_system_reference())
async def _create_directory_and_return_client(self, directory=None):
directory_name = directory if directory else self._get_directory_reference()
directory_client = self.dsc.get_directory_client(self.file_system_name, directory_name)
await directory_client.create_directory()
return directory_client
async def _create_file_and_return_client(self, directory="", file=None):
if directory:
await self._create_directory_and_return_client(directory)
if not file:
file = self._get_file_reference()
file_client = self.dsc.get_file_client(self.file_system_name, directory + '/' + file)
await file_client.create_file()
return file_client
# --Helpers-----------------------------------------------------------------
@DataLakePreparer()
async def test_create_file_async(self, datalake_storage_account_name, datalake_storage_account_key):
await self._setUp(datalake_storage_account_name, datalake_storage_account_key)
# Arrange
directory_name = self._get_directory_reference()
# Create a directory to put the file under that
directory_client = self.dsc.get_directory_client(self.file_system_name, directory_name)
await directory_client.create_directory()
file_client = directory_client.get_file_client('filename')
response = await file_client.create_file()
# Assert
self.assertIsNotNone(response)
@DataLakePreparer()
async def test_file_exists(self, datalake_storage_account_name, datalake_storage_account_key):
await self._setUp(datalake_storage_account_name, datalake_storage_account_key)
# Arrange
directory_name = self._get_directory_reference()
directory_client = self.dsc.get_directory_client(self.file_system_name, directory_name)
await directory_client.create_directory()
file_client1 = directory_client.get_file_client('filename')
file_client2 = directory_client.get_file_client('nonexistentfile')
await file_client1.create_file()
self.assertTrue(await file_client1.exists())
self.assertFalse(await file_client2.exists())
@DataLakePreparer()
async def test_create_file_using_oauth_token_credential_async(self, datalake_storage_account_name, datalake_storage_account_key):
await self._setUp(datalake_storage_account_name, datalake_storage_account_key)
# Arrange
file_name = self._get_file_reference()
token_credential = self.generate_oauth_token()
# Create a directory to put the file under that
file_client = DataLakeFileClient(self.dsc.url, self.file_system_name, file_name,
credential=token_credential)
response = await file_client.create_file()
# Assert
self.assertIsNotNone(response)
@DataLakePreparer()
async def test_create_file_with_existing_name_async(self, datalake_storage_account_name, datalake_storage_account_key):
await self._setUp(datalake_storage_account_name, datalake_storage_account_key)
# Arrange
file_client = await self._create_file_and_return_client()
with self.assertRaises(ResourceExistsError):
# if the file exists then throw error
# if_none_match='*' is to make sure no existing file
await file_client.create_file(match_condition=MatchConditions.IfMissing)
@DataLakePreparer()
async def test_create_file_with_lease_id_async(self, datalake_storage_account_name, datalake_storage_account_key):
await self._setUp(datalake_storage_account_name, datalake_storage_account_key)
# Arrange
directory_name = self._get_directory_reference()
directory_client = self.dsc.get_directory_client(self.file_system_name, directory_name)
await directory_client.create_directory()
file_client = directory_client.get_file_client('filename')
# Act
await file_client.create_file()
lease = await file_client.acquire_lease()
create_resp = await file_client.create_file(lease=lease)
# Assert
file_properties = await file_client.get_file_properties()
self.assertIsNotNone(file_properties)
self.assertEqual(file_properties.etag, create_resp.get('etag'))
self.assertEqual(file_properties.last_modified, create_resp.get('last_modified'))
@DataLakePreparer()
async def test_create_file_under_root_directory_async(self, datalake_storage_account_name, datalake_storage_account_key):
await self._setUp(datalake_storage_account_name, datalake_storage_account_key)
# Arrange
# get a file client to interact with the file under root directory
file_client = self.dsc.get_file_client(self.file_system_name, "filename")
response = await file_client.create_file()
# Assert
self.assertIsNotNone(response)
@DataLakePreparer()
async def test_append_data_async(self, datalake_storage_account_name, datalake_storage_account_key):
await self._setUp(datalake_storage_account_name, datalake_storage_account_key)
directory_name = self._get_directory_reference()
# Create a directory to put the file under that
directory_client = self.dsc.get_directory_client(self.file_system_name, directory_name)
await directory_client.create_directory()
file_client = directory_client.get_file_client('filename')
await file_client.create_file()
# Act
response = await file_client.append_data(b'abc', 0, 3)
self.assertIsNotNone(response)
@DataLakePreparer()
async def test_append_empty_data_async(self, datalake_storage_account_name, datalake_storage_account_key):
await self._setUp(datalake_storage_account_name, datalake_storage_account_key)
file_client = await self._create_file_and_return_client()
# Act
await file_client.flush_data(0)
file_props = await file_client.get_file_properties()
self.assertIsNotNone(file_props['size'], 0)
@DataLakePreparer()
async def test_flush_data_async(self, datalake_storage_account_name, datalake_storage_account_key):
await self._setUp(datalake_storage_account_name, datalake_storage_account_key)
directory_name = self._get_directory_reference()
# Create a directory to put the file under that
directory_client = self.dsc.get_directory_client(self.file_system_name, directory_name)
await directory_client.create_directory()
file_client = directory_client.get_file_client('filename')
await file_client.create_file()
# Act
await file_client.append_data(b'abc', 0, 3)
response = await file_client.flush_data(3)
# Assert
prop = await file_client.get_file_properties()
self.assertIsNotNone(response)
self.assertEqual(prop['size'], 3)
@DataLakePreparer()
async def test_flush_data_with_match_condition_async(self, datalake_storage_account_name, datalake_storage_account_key):
await self._setUp(datalake_storage_account_name, datalake_storage_account_key)
directory_name = self._get_directory_reference()
# Create a directory to put the file under that
directory_client = self.dsc.get_directory_client(self.file_system_name, directory_name)
await directory_client.create_directory()
file_client = directory_client.get_file_client('filename')
resp = await file_client.create_file()
# Act
await file_client.append_data(b'abc', 0, 3)
# flush is successful because it isn't touched
response = await file_client.flush_data(3, etag=resp['etag'], match_condition=MatchConditions.IfNotModified)
await file_client.append_data(b'abc', 3, 3)
with self.assertRaises(ResourceModifiedError):
# flush is unsuccessful because extra data were appended.
await file_client.flush_data(6, etag=resp['etag'], match_condition=MatchConditions.IfNotModified)
@pytest.mark.live_test_only
@DataLakePreparer()
async def test_upload_data_in_substreams(self, datalake_storage_account_name, datalake_storage_account_key):
await self._setUp(datalake_storage_account_name, datalake_storage_account_key)
# parallel upload cannot be recorded
directory_name = self._get_directory_reference()
# Create a directory to put the file under that
directory_client = self.dsc.get_directory_client(self.file_system_name, directory_name)
await directory_client.create_directory()
file_client = directory_client.get_file_client('filename')
# Get 16MB data
raw_data = self.get_random_bytes(16 * 1024 * 1024)
# Ensure chunk size is greater than threshold (8MB > 4MB) - for optimized upload
await file_client.upload_data(raw_data, chunk_size=8 * 1024 * 1024, overwrite=True, max_concurrency=3)
data = await file_client.download_file()
downloaded_data = await data.readall()
self.assertEqual(raw_data, downloaded_data)
# Run on single thread
await file_client.upload_data(raw_data, chunk_size=8 * 1024 * 1024, overwrite=True)
data = await file_client.download_file()
downloaded_data = await data.readall()
self.assertEqual(raw_data, downloaded_data)
@pytest.mark.live_test_only
@DataLakePreparer()
async def test_upload_data_async(self, datalake_storage_account_name, datalake_storage_account_key):
await self._setUp(datalake_storage_account_name, datalake_storage_account_key)
# parallel upload cannot be recorded
directory_name = self._get_directory_reference()
# Create a directory to put the file under that
directory_client = self.dsc.get_directory_client(self.file_system_name, directory_name)
await directory_client.create_directory()
file_client = directory_client.get_file_client('filename')
data = self.get_random_bytes(400*1024)
await file_client.upload_data(data, overwrite=True, max_concurrency=5)
downloaded_data = await (await file_client.download_file()).readall()
self.assertEqual(data, downloaded_data)
@pytest.mark.live_test_only
@DataLakePreparer()
async def test_upload_data_to_existing_file_async(self, datalake_storage_account_name, datalake_storage_account_key):
await self._setUp(datalake_storage_account_name, datalake_storage_account_key)
directory_name = self._get_directory_reference()
# Create a directory to put the file under that
directory_client = self.dsc.get_directory_client(self.file_system_name, directory_name)
await directory_client.create_directory()
# create an existing file
file_client = directory_client.get_file_client('filename')
await file_client.create_file()
await file_client.append_data(b"abc", 0)
await file_client.flush_data(3)
# to override the existing file
data = self.get_random_bytes(100)
with self.assertRaises(HttpResponseError):
await file_client.upload_data(data, max_concurrency=5)
await file_client.upload_data(data, overwrite=True, max_concurrency=5)
downloaded_data = await (await file_client.download_file()).readall()
self.assertEqual(data, downloaded_data)
@pytest.mark.live_test_only
@DataLakePreparer()
async def test_upload_data_to_existing_file_with_content_settings_async(self, datalake_storage_account_name, datalake_storage_account_key):
await self._setUp(datalake_storage_account_name, datalake_storage_account_key)
# etag in async recording file cannot be parsed properly
directory_name = self._get_directory_reference()
# Create a directory to put the file under that
directory_client = self.dsc.get_directory_client(self.file_system_name, directory_name)
await directory_client.create_directory()
# create an existing file
file_client = directory_client.get_file_client('filename')
resp = await file_client.create_file()
etag = resp['etag']
# to override the existing file
data = self.get_random_bytes(100)
content_settings = ContentSettings(
content_language='spanish',
content_disposition='inline')
await file_client.upload_data(data, max_concurrency=5,
content_settings=content_settings, etag=etag,
match_condition=MatchConditions.IfNotModified)
downloaded_data = await (await file_client.download_file()).readall()
properties = await file_client.get_file_properties()
self.assertEqual(data, downloaded_data)
self.assertEqual(properties.content_settings.content_language, content_settings.content_language)
@pytest.mark.live_test_only
@DataLakePreparer()
async def test_upload_data_to_existing_file_with_permissions_and_umask_async(self, datalake_storage_account_name, datalake_storage_account_key):
await self._setUp(datalake_storage_account_name, datalake_storage_account_key)
# etag in async recording file cannot be parsed properly
directory_name = self._get_directory_reference()
# Create a directory to put the file under that
directory_client = self.dsc.get_directory_client(self.file_system_name, directory_name)
await directory_client.create_directory()
# create an existing file
file_client = directory_client.get_file_client('filename')
resp = await file_client.create_file()
etag = resp['etag']
# to override the existing file
data = self.get_random_bytes(100)
await file_client.upload_data(data,
overwrite=True, max_concurrency=5,
permissions='0777', umask="0000",
etag=etag,
match_condition=MatchConditions.IfNotModified)
downloaded_data = await (await file_client.download_file()).readall()
prop = await file_client.get_access_control()
self.assertEqual(data, downloaded_data)
self.assertEqual(prop['permissions'], 'rwxrwxrwx')
@DataLakePreparer()
async def test_read_file_async(self, datalake_storage_account_name, datalake_storage_account_key):
await self._setUp(datalake_storage_account_name, datalake_storage_account_key)
file_client = await self._create_file_and_return_client()
data = self.get_random_bytes(1024)
# upload data to file
await file_client.append_data(data, 0, len(data))
await file_client.flush_data(len(data))
# doanload the data and make sure it is the same as uploaded data
downloaded_data = await (await file_client.download_file()).readall()
self.assertEqual(data, downloaded_data)
@pytest.mark.live_test_only
@DataLakePreparer()
async def test_read_file_with_user_delegation_key_async(self, datalake_storage_account_name, datalake_storage_account_key):
await self._setUp(datalake_storage_account_name, datalake_storage_account_key)
# SAS URL is calculated from storage key, so this test runs live only
# Create file
file_client = await self._create_file_and_return_client()
data = self.get_random_bytes(1024)
# Upload data to file
await file_client.append_data(data, 0, len(data))
await file_client.flush_data(len(data))
# Get user delegation key
token_credential = self.generate_oauth_token()
service_client = DataLakeServiceClient(self.account_url(datalake_storage_account_name, 'dfs'), credential=token_credential)
user_delegation_key = await service_client.get_user_delegation_key(datetime.utcnow(),
datetime.utcnow() + timedelta(hours=1))
sas_token = generate_file_sas(file_client.account_name,
file_client.file_system_name,
None,
file_client.path_name,
user_delegation_key,
permission=FileSasPermissions(read=True, create=True, write=True, delete=True),
expiry=datetime.utcnow() + timedelta(hours=1),
)
# doanload the data and make sure it is the same as uploaded data
new_file_client = DataLakeFileClient(self.account_url(datalake_storage_account_name, 'dfs'),
file_client.file_system_name,
file_client.path_name,
credential=sas_token)
downloaded_data = await (await new_file_client.download_file()).readall()
self.assertEqual(data, downloaded_data)
@DataLakePreparer()
async def test_read_file_into_file_async(self, datalake_storage_account_name, datalake_storage_account_key):
await self._setUp(datalake_storage_account_name, datalake_storage_account_key)
file_client = await self._create_file_and_return_client()
data = self.get_random_bytes(1024)
# upload data to file
await file_client.append_data(data, 0, len(data))
await file_client.flush_data(len(data))
# doanload the data into a file and make sure it is the same as uploaded data
with open(FILE_PATH, 'wb') as stream:
download = await file_client.download_file(max_concurrency=2)
await download.readinto(stream)
# Assert
with open(FILE_PATH, 'rb') as stream:
actual = stream.read()
self.assertEqual(data, actual)
@DataLakePreparer()
async def test_read_file_to_text_async(self, datalake_storage_account_name, datalake_storage_account_key):
await self._setUp(datalake_storage_account_name, datalake_storage_account_key)
file_client = await self._create_file_and_return_client()
data = self.get_random_text_data(1024)
# upload data to file
await file_client.append_data(data, 0, len(data))
await file_client.flush_data(len(data))
# doanload the text data and make sure it is the same as uploaded data
downloaded_data = await (await file_client.download_file(max_concurrency=2, encoding="utf-8")).readall()
# Assert
self.assertEqual(data, downloaded_data)
@pytest.mark.live_test_only
@DataLakePreparer()
async def test_account_sas_async(self, datalake_storage_account_name, datalake_storage_account_key):
await self._setUp(datalake_storage_account_name, datalake_storage_account_key)
# SAS URL is calculated from storage key, so this test runs live only
file_name = self._get_file_reference()
# create a file under root directory
await self._create_file_and_return_client(file=file_name)
# generate a token with file level read permission
token = generate_account_sas(
self.dsc.account_name,
self.dsc.credential.account_key,
ResourceTypes(file_system=True, object=True),
AccountSasPermissions(read=True),
datetime.utcnow() + timedelta(hours=1),
)
for credential in [token, AzureSasCredential(token)]:
# read the created file which is under root directory
file_client = DataLakeFileClient(self.dsc.url, self.file_system_name, file_name, credential=credential)
properties = await file_client.get_file_properties()
# make sure we can read the file properties
self.assertIsNotNone(properties)
# try to write to the created file with the token
with self.assertRaises(HttpResponseError):
await file_client.append_data(b"abcd", 0, 4)
@DataLakePreparer()
async def test_account_sas_raises_if_sas_already_in_uri(self, datalake_storage_account_name, datalake_storage_account_key):
await self._setUp(datalake_storage_account_name, datalake_storage_account_key)
with self.assertRaises(ValueError):
DataLakeFileClient(self.dsc.url + "?sig=foo", self.file_system_name, "foo", credential=AzureSasCredential("?foo=bar"))
@pytest.mark.live_test_only
@DataLakePreparer()
async def test_file_sas_only_applies_to_file_level_async(self, datalake_storage_account_name, datalake_storage_account_key):
await self._setUp(datalake_storage_account_name, datalake_storage_account_key)
# SAS URL is calculated from storage key, so this test runs live only
file_name = self._get_file_reference()
directory_name = self._get_directory_reference()
await self._create_file_and_return_client(directory=directory_name, file=file_name)
# generate a token with file level read and write permissions
token = generate_file_sas(
self.dsc.account_name,
self.file_system_name,
directory_name,
file_name,
self.dsc.credential.account_key,
permission=FileSasPermissions(read=True, write=True),
expiry=datetime.utcnow() + timedelta(hours=1),
)
# read the created file which is under root directory
file_client = DataLakeFileClient(self.dsc.url, self.file_system_name, directory_name + '/' + file_name,
credential=token)
properties = await file_client.get_file_properties()
# make sure we can read the file properties
self.assertIsNotNone(properties)
# try to write to the created file with the token
response = await file_client.append_data(b"abcd", 0, 4, validate_content=True)
self.assertIsNotNone(response)
# the token is for file level, so users are not supposed to have access to file system level operations
file_system_client = FileSystemClient(self.dsc.url, self.file_system_name, credential=token)
with self.assertRaises(ClientAuthenticationError):
await file_system_client.get_file_system_properties()
# the token is for file level, so users are not supposed to have access to directory level operations
directory_client = DataLakeDirectoryClient(self.dsc.url, self.file_system_name, directory_name,
credential=token)
with self.assertRaises(ClientAuthenticationError):
await directory_client.get_directory_properties()
@DataLakePreparer()
async def test_delete_file_async(self, datalake_storage_account_name, datalake_storage_account_key):
await self._setUp(datalake_storage_account_name, datalake_storage_account_key)
# Arrange
file_client = await self._create_file_and_return_client()
await file_client.delete_file()
with self.assertRaises(ResourceNotFoundError):
await file_client.get_file_properties()
@DataLakePreparer()
async def test_delete_file_with_if_unmodified_since_async(self, datalake_storage_account_name, datalake_storage_account_key):
await self._setUp(datalake_storage_account_name, datalake_storage_account_key)
# Arrange
file_client = await self._create_file_and_return_client()
prop = await file_client.get_file_properties()
await file_client.delete_file(if_unmodified_since=prop['last_modified'])
# Make sure the file was deleted
with self.assertRaises(ResourceNotFoundError):
await file_client.get_file_properties()
@DataLakePreparer()
async def test_set_access_control_async(self, datalake_storage_account_name, datalake_storage_account_key):
await self._setUp(datalake_storage_account_name, datalake_storage_account_key)
file_client = await self._create_file_and_return_client()
response = await file_client.set_access_control(permissions='0777')
# Assert
self.assertIsNotNone(response)
@DataLakePreparer()
async def test_set_access_control_with_match_conditions_async(self, datalake_storage_account_name, datalake_storage_account_key):
await self._setUp(datalake_storage_account_name, datalake_storage_account_key)
file_client = await self._create_file_and_return_client()
with self.assertRaises(ResourceModifiedError):
await file_client.set_access_control(permissions='0777', match_condition=MatchConditions.IfMissing)
@DataLakePreparer()
async def test_get_access_control_async(self, datalake_storage_account_name, datalake_storage_account_key):
await self._setUp(datalake_storage_account_name, datalake_storage_account_key)
file_client = await self._create_file_and_return_client()
await file_client.set_access_control(permissions='0777')
# Act
response = await file_client.get_access_control()
# Assert
self.assertIsNotNone(response)
@DataLakePreparer()
async def test_get_access_control_with_if_modified_since_async(self, datalake_storage_account_name, datalake_storage_account_key):
await self._setUp(datalake_storage_account_name, datalake_storage_account_key)
file_client = await self._create_file_and_return_client()
await file_client.set_access_control(permissions='0777')
prop = await file_client.get_file_properties()
# Act
response = await file_client.get_access_control(if_modified_since=prop['last_modified'] - timedelta(minutes=15))
# Assert
self.assertIsNotNone(response)
@DataLakePreparer()
async def test_get_properties_async(self, datalake_storage_account_name, datalake_storage_account_key):
await self._setUp(datalake_storage_account_name, datalake_storage_account_key)
# Arrange
directory_client = await self._create_directory_and_return_client()
metadata = {'hello': 'world', 'number': '42'}
content_settings = ContentSettings(
content_language='spanish',
content_disposition='inline')
file_client = await directory_client.create_file("newfile", metadata=metadata,
content_settings=content_settings)
await file_client.append_data(b"abc", 0, 3)
await file_client.flush_data(3)
properties = await file_client.get_file_properties()
# Assert
self.assertTrue(properties)
self.assertEqual(properties.size, 3)
self.assertEqual(properties.metadata['hello'], metadata['hello'])
self.assertEqual(properties.content_settings.content_language, content_settings.content_language)
@DataLakePreparer()
async def test_set_access_control_recursive_async(self, datalake_storage_account_name, datalake_storage_account_key):
await self._setUp(datalake_storage_account_name, datalake_storage_account_key)
acl = 'user::rwx,group::r-x,other::rwx'
file_client = await self._create_file_and_return_client()
summary = await file_client.set_access_control_recursive(acl=acl)
# Assert
self.assertEqual(summary.counters.directories_successful, 0)
self.assertEqual(summary.counters.files_successful, 1)
self.assertEqual(summary.counters.failure_count, 0)
access_control = await file_client.get_access_control()
self.assertIsNotNone(access_control)
self.assertEqual(acl, access_control['acl'])
@DataLakePreparer()
async def test_update_access_control_recursive_async(self, datalake_storage_account_name, datalake_storage_account_key):
await self._setUp(datalake_storage_account_name, datalake_storage_account_key)
acl = 'user::rwx,group::r-x,other::rwx'
file_client = await self._create_file_and_return_client()
summary = await file_client.update_access_control_recursive(acl=acl)
# Assert
self.assertEqual(summary.counters.directories_successful, 0)
self.assertEqual(summary.counters.files_successful, 1)
self.assertEqual(summary.counters.failure_count, 0)
access_control = await file_client.get_access_control()
self.assertIsNotNone(access_control)
self.assertEqual(acl, access_control['acl'])
@DataLakePreparer()
async def test_remove_access_control_recursive_async(self, datalake_storage_account_name, datalake_storage_account_key):
await self._setUp(datalake_storage_account_name, datalake_storage_account_key)
acl = "mask," + "default:user,default:group," + \
"user:ec3595d6-2c17-4696-8caa-7e139758d24a,group:ec3595d6-2c17-4696-8caa-7e139758d24a," + \
"default:user:ec3595d6-2c17-4696-8caa-7e139758d24a,default:group:ec3595d6-2c17-4696-8caa-7e139758d24a"
file_client = await self._create_file_and_return_client()
summary = await file_client.remove_access_control_recursive(acl=acl)
# Assert
self.assertEqual(summary.counters.directories_successful, 0)
self.assertEqual(summary.counters.files_successful, 1)
self.assertEqual(summary.counters.failure_count, 0)
@DataLakePreparer()
async def test_set_expiry_async(self, datalake_storage_account_name, datalake_storage_account_key):
await self._setUp(datalake_storage_account_name, datalake_storage_account_key)
# Arrange
directory_client = await self._create_directory_and_return_client()
metadata = {'hello': 'world', 'number': '42'}
content_settings = ContentSettings(
content_language='spanish',
content_disposition='inline')
expires_on = datetime.utcnow() + timedelta(hours=1)
file_client = await directory_client.create_file("newfile", metadata=metadata, content_settings=content_settings)
await file_client.set_file_expiry("Absolute", expires_on=expires_on)
properties = await file_client.get_file_properties()
# Assert
self.assertTrue(properties)
self.assertIsNotNone(properties.expiry_time)
@DataLakePreparer()
async def test_rename_file_with_non_used_name_async(self, datalake_storage_account_name, datalake_storage_account_key):
await self._setUp(datalake_storage_account_name, datalake_storage_account_key)
file_client = await self._create_file_and_return_client()
data_bytes = b"abc"
await file_client.append_data(data_bytes, 0, 3)
await file_client.flush_data(3)
new_client = await file_client.rename_file(file_client.file_system_name + '/' + 'newname')
data = await (await new_client.download_file()).readall()
self.assertEqual(data, data_bytes)
self.assertEqual(new_client.path_name, "newname")
@DataLakePreparer()
async def test_rename_file_to_existing_file_async(self, datalake_storage_account_name, datalake_storage_account_key):
await self._setUp(datalake_storage_account_name, datalake_storage_account_key)
# create the existing file
existing_file_client = await self._create_file_and_return_client(file="existingfile")
await existing_file_client.append_data(b"a", 0, 1)
await existing_file_client.flush_data(1)
old_url = existing_file_client.url
# prepare to rename the file to the existing file
file_client = await self._create_file_and_return_client()
data_bytes = b"abc"
await file_client.append_data(data_bytes, 0, 3)
await file_client.flush_data(3)
new_client = await file_client.rename_file(file_client.file_system_name + '/' + existing_file_client.path_name)
new_url = file_client.url
data = await (await new_client.download_file()).readall()
# the existing file was overridden
self.assertEqual(data, data_bytes)
@pytest.mark.live_test_only
@DataLakePreparer()
async def test_rename_file_with_file_sas_async(self, datalake_storage_account_name, datalake_storage_account_key):
await self._setUp(datalake_storage_account_name, datalake_storage_account_key)
# SAS URL is calculated from storage key, so this test runs live only
token = generate_file_sas(self.dsc.account_name,
self.file_system_name,
None,
"oldfile",
datalake_storage_account_key,
permission=FileSasPermissions(read=True, create=True, write=True, delete=True, move=True),
expiry=datetime.utcnow() + timedelta(hours=1),
)
new_token = generate_file_sas(self.dsc.account_name,
self.file_system_name,
None,
"newname",
datalake_storage_account_key,
permission=FileSasPermissions(read=True, create=True, write=True, delete=True),
expiry=datetime.utcnow() + timedelta(hours=1),
)
# read the created file which is under root directory
file_client = DataLakeFileClient(self.dsc.url, self.file_system_name, "oldfile", credential=token)
await file_client.create_file()
data_bytes = b"abc"
await file_client.append_data(data_bytes, 0, 3)
await file_client.flush_data(3)
new_client = await file_client.rename_file(file_client.file_system_name+'/'+'newname'+'?'+new_token)
data = await (await new_client.download_file()).readall()
self.assertEqual(data, data_bytes)
self.assertEqual(new_client.path_name, "newname")
@pytest.mark.live_test_only
@DataLakePreparer()
async def test_rename_file_will_not_change_existing_directory_async(self, datalake_storage_account_name, datalake_storage_account_key):
await self._setUp(datalake_storage_account_name, datalake_storage_account_key)
# create none empty directory(with 2 files)
dir1 = await self._create_directory_and_return_client(directory="dir1")
f1 = await dir1.create_file("file1")
await f1.append_data(b"file1", 0, 5)
await f1.flush_data(5)
f2 = await dir1.create_file("file2")
await f2.append_data(b"file2", 0, 5)
await f2.flush_data(5)
# create another none empty directory(with 2 files)
dir2 = await self._create_directory_and_return_client(directory="dir2")
f3 = await dir2.create_file("file3")
await f3.append_data(b"file3", 0, 5)
await f3.flush_data(5)
f4 = await dir2.create_file("file4")
await f4.append_data(b"file4", 0, 5)
await f4.flush_data(5)
new_client = await f3.rename_file(f1.file_system_name + '/' + f1.path_name)
self.assertEqual(await (await new_client.download_file()).readall(), b"file3")
# make sure the data in file2 and file4 weren't touched
f2_data = await (await f2.download_file()).readall()
self.assertEqual(f2_data, b"file2")
f4_data = await (await f4.download_file()).readall()
self.assertEqual(f4_data, b"file4")
with self.assertRaises(HttpResponseError):
await (await f3.download_file()).readall()
class FileSystemTest(StorageTestCase):
def setUp(self):
super(FileSystemTest, self).setUp()
url = self._get_account_url()
self.dsc = DataLakeServiceClient(
url,
credential=self.settings.STORAGE_DATA_LAKE_ACCOUNT_KEY,
transport=AiohttpTestTransport())
self.config = self.dsc._config
loop = asyncio.get_event_loop()
loop.run_until_complete(self.dsc.__aenter__())
self.test_file_systems = []
def tearDown(self):
if not self.is_playback():
loop = asyncio.get_event_loop()
try:
for file_system in self.test_file_systems:
loop.run_until_complete(
self.dsc.delete_file_system(file_system))
loop.run_until_complete(self.fsc.__aexit__())
except:
pass
return super(FileSystemTest, self).tearDown()
# --Helpers-----------------------------------------------------------------
def _get_file_system_reference(self, prefix=TEST_FILE_SYSTEM_PREFIX):
file_system_name = self.get_resource_name(prefix)
self.test_file_systems.append(file_system_name)
return file_system_name
async def _create_file_system(self,
file_system_prefix=TEST_FILE_SYSTEM_PREFIX):
return await self.dsc.create_file_system(
self._get_file_system_reference(prefix=file_system_prefix))
# --Helpers-----------------------------------------------------------------
async def _test_create_file_system_async(self):
# Arrange
file_system_name = self._get_file_system_reference()
# Act
file_system_client = self.dsc.get_file_system_client(file_system_name)
created = await file_system_client.create_file_system()
# Assert
self.assertTrue(created)
@record
def test_create_file_system_async(self):
loop = asyncio.get_event_loop()
loop.run_until_complete(self._test_create_file_system_async())
async def _test_create_file_system_with_metadata_async(self):
# Arrange
metadata = {'hello': 'world', 'number': '42'}
file_system_name = self._get_file_system_reference()
# Act
file_system_client = self.dsc.get_file_system_client(file_system_name)
created = await file_system_client.create_file_system(metadata=metadata
)
# Assert
properties = await file_system_client.get_file_system_properties()
self.assertTrue(created)
self.assertDictEqual(properties.metadata, metadata)
@record
def test_create_file_system_with_metadata_async(self):
loop = asyncio.get_event_loop()
loop.run_until_complete(
self._test_create_file_system_with_metadata_async())
async def _test_list_file_systems_async(self):
# Arrange
file_system_name = self._get_file_system_reference()
file_system = await self.dsc.create_file_system(file_system_name)
# Act
file_systems = []
async for filesystem in self.dsc.list_file_systems():
file_systems.append(filesystem)
# Assert
self.assertIsNotNone(file_systems)
self.assertGreaterEqual(len(file_systems), 1)
self.assertIsNotNone(file_systems[0])
self.assertNamedItemInContainer(file_systems,
file_system.file_system_name)
self.assertIsNotNone(file_systems[0].has_immutability_policy)
self.assertIsNotNone(file_systems[0].has_legal_hold)
@record
def test_list_file_systems_async(self):
loop = asyncio.get_event_loop()
loop.run_until_complete(self._test_list_file_systems_async())
async def _test_delete_file_system_with_existing_file_system_async(self):
# Arrange
file_system = await self._create_file_system()
# Act
deleted = await file_system.delete_file_system()
# Assert
self.assertIsNone(deleted)
@record
def test_delete_file_system_with_existing_file_system_async(self):
loop = asyncio.get_event_loop()
loop.run_until_complete(
self._test_delete_file_system_with_existing_file_system_async())
async def _test_delete_none_existing_file_system_async(self):
fake_file_system_client = self.dsc.get_file_system_client("fakeclient")
# Act
with self.assertRaises(ResourceNotFoundError):
await fake_file_system_client.delete_file_system(
match_condition=MatchConditions.IfMissing)
@record
def test_delete_none_existing_file_system_async(self):
loop = asyncio.get_event_loop()
loop.run_until_complete(
self._test_delete_none_existing_file_system_async())
async def _test_list_file_systems_with_include_metadata_async(self):
# Arrange
file_system = await self._create_file_system()
metadata = {'hello': 'world', 'number': '42'}
await file_system.set_file_system_metadata(metadata)
# Act
file_systems = []
async for fs in self.dsc.list_file_systems(
name_starts_with=file_system.file_system_name,
include_metadata=True):
file_systems.append(fs)
# Assert
self.assertIsNotNone(file_systems)
self.assertGreaterEqual(len(file_systems), 1)
self.assertIsNotNone(file_systems[0])
self.assertNamedItemInContainer(file_systems,
file_system.file_system_name)
self.assertDictEqual(file_systems[0].metadata, metadata)
@record
def test_list_file_systems_with_include_metadata_async(self):
loop = asyncio.get_event_loop()
loop.run_until_complete(
self._test_list_file_systems_with_include_metadata_async())
async def _test_set_file_system_acl_async(self):
# Act
file_system = await self._create_file_system()
access_policy = AccessPolicy(
permission=FileSystemSasPermissions(read=True),
expiry=datetime.utcnow() + timedelta(hours=1),
start=datetime.utcnow())
signed_identifier1 = {'testid': access_policy}
response = await file_system.set_file_system_access_policy(
signed_identifier1, public_access=PublicAccess.FileSystem)
self.assertIsNotNone(response.get('etag'))
self.assertIsNotNone(response.get('last_modified'))
acl1 = await file_system.get_file_system_access_policy()
self.assertIsNotNone(acl1['public_access'])
self.assertEqual(len(acl1['signed_identifiers']), 1)
# If set signed identifier without specifying the access policy then it will be default to None
signed_identifier2 = {'testid': access_policy, 'test2': access_policy}
await file_system.set_file_system_access_policy(signed_identifier2)
acl2 = await file_system.get_file_system_access_policy()
self.assertIsNone(acl2['public_access'])
self.assertEqual(len(acl2['signed_identifiers']), 2)
@record
def test_set_file_system_acl_async(self):
loop = asyncio.get_event_loop()
loop.run_until_complete(self._test_set_file_system_acl_async())
async def _test_list_file_systems_by_page_async(self):
# Arrange
for i in range(0, 6):
await self._create_file_system(
file_system_prefix="filesystem{}".format(i))
# Act
file_systems = []
async for fs in await self.dsc.list_file_systems(
results_per_page=3, name_starts_with="file",
include_metadata=True).by_page().__anext__():
file_systems.append(fs)
# Assert
self.assertIsNotNone(file_systems)
self.assertGreaterEqual(len(file_systems), 3)
@record
def test_list_file_systems_by_page_async(self):
loop = asyncio.get_event_loop()
loop.run_until_complete(self._test_list_file_systems_by_page_async())
async def _test_list_file_systems_with_public_access_async(self):
# Arrange
file_system_name = self._get_file_system_reference()
file_system = self.dsc.get_file_system_client(file_system_name)
await file_system.create_file_system(public_access="blob")
metadata = {'hello': 'world', 'number': '42'}
await file_system.set_file_system_metadata(metadata)
# Act
file_systems = []
async for fs in self.dsc.list_file_systems(
name_starts_with=file_system.file_system_name,
include_metadata=True):
file_systems.append(fs)
# Assert
self.assertIsNotNone(file_systems)
self.assertGreaterEqual(len(file_systems), 1)
self.assertIsNotNone(file_systems[0])
self.assertNamedItemInContainer(file_systems,
file_system.file_system_name)
self.assertDictEqual(file_systems[0].metadata, metadata)
self.assertTrue(file_systems[0].public_access is PublicAccess.File)
@record
def test_list_file_systems_with_public_access_async(self):
loop = asyncio.get_event_loop()
loop.run_until_complete(
self._test_list_file_systems_with_public_access_async())
async def _test_get_file_system_properties_async(self):
# Arrange
metadata = {'hello': 'world', 'number': '42'}
file_system = await self._create_file_system()
await file_system.set_file_system_metadata(metadata)
# Act
props = await file_system.get_file_system_properties()
# Assert
self.assertIsNotNone(props)
self.assertDictEqual(props.metadata, metadata)
self.assertIsNotNone(props.has_immutability_policy)
self.assertIsNotNone(props.has_legal_hold)
@record
def test_get_file_system_properties_async(self):
loop = asyncio.get_event_loop()
loop.run_until_complete(self._test_get_file_system_properties_async())
async def _test_list_paths_async(self):
# Arrange
file_system = await self._create_file_system()
for i in range(0, 6):
await file_system.create_directory("dir1{}".format(i))
paths = []
async for path in file_system.get_paths(upn=True):
paths.append(path)
self.assertEqual(len(paths), 6)
@record
def test_list_paths_async(self):
loop = asyncio.get_event_loop()
loop.run_until_complete(self._test_list_paths_async())
async def _test_list_paths_which_are_all_files_async(self):
# Arrange
file_system = await self._create_file_system()
for i in range(0, 6):
await file_system.create_file("file{}".format(i))
paths = []
async for path in file_system.get_paths(upn=True):
paths.append(path)
self.assertEqual(len(paths), 6)
@record
def test_list_paths_which_are_all_files_async(self):
loop = asyncio.get_event_loop()
loop.run_until_complete(
self._test_list_paths_which_are_all_files_async())
async def _test_list_paths_with_max_per_page_async(self):
# Arrange
file_system = await self._create_file_system()
for i in range(0, 6):
await file_system.create_directory("dir1{}".format(i))
generator1 = file_system.get_paths(max_results=2, upn=True).by_page()
paths1 = []
async for path in await generator1.__anext__():
paths1.append(path)
generator2 = file_system.get_paths(max_results=4, upn=True) \
.by_page(continuation_token=generator1.continuation_token)
paths2 = []
async for path in await generator2.__anext__():
paths2.append(path)
self.assertEqual(len(paths1), 2)
self.assertEqual(len(paths2), 4)
@record
def test_list_paths_with_max_per_page_async(self):
loop = asyncio.get_event_loop()
loop.run_until_complete(
self._test_list_paths_with_max_per_page_async())
async def _test_list_paths_under_specific_path_async(self):
# Arrange
file_system = await self._create_file_system()
for i in range(0, 6):
dir = await file_system.create_directory("dir1{}".format(i))
# create a subdirectory under the current directory
subdir = await dir.create_sub_directory("subdir")
await subdir.create_sub_directory("subsub")
# create a file under the current directory
file_client = await subdir.create_file("file")
await file_client.append_data(b"abced", 0, 5)
await file_client.flush_data(5)
generator1 = file_system.get_paths(path="dir10/subdir",
max_results=2,
upn=True).by_page()
paths = []
async for path in await generator1.__anext__():
paths.append(path)
self.assertEqual(len(paths), 2)
self.assertEqual(paths[0].content_length, 5)
@record
def test_list_paths_under_specific_path_async(self):
loop = asyncio.get_event_loop()
loop.run_until_complete(
self._test_list_paths_under_specific_path_async())
async def _test_list_paths_recursively_async(self):
# Arrange
file_system = await self._create_file_system()
for i in range(0, 6):
dir = await file_system.create_directory("dir1{}".format(i))
# create a subdirectory under the current directory
subdir = await dir.create_sub_directory("subdir")
await subdir.create_sub_directory("subsub")
# create a file under the current directory
await subdir.create_file("file")
paths = []
async for path in file_system.get_paths(recursive=True, upn=True):
paths.append(path)
# there are 24 subpaths in total
self.assertEqual(len(paths), 24)
@record
def test_list_paths_recursively_async(self):
loop = asyncio.get_event_loop()
loop.run_until_complete(self._test_list_paths_recursively_async())
async def _test_create_directory_from_file_system_client_async(self):
# Arrange
file_system = await self._create_file_system()
await file_system.create_directory("dir1/dir2")
paths = []
async for path in file_system.get_paths(recursive=False, upn=True):
paths.append(path)
self.assertEqual(len(paths), 1)
self.assertEqual(paths[0].name, "dir1")
@record
def test_create_directory_from_file_system_client_async(self):
loop = asyncio.get_event_loop()
loop.run_until_complete(
self._test_create_directory_from_file_system_client_async())
async def _test_create_file_from_file_system_client_async(self):
# Arrange
file_system = await self._create_file_system()
await file_system.create_file("dir1/dir2/file")
paths = []
async for path in file_system.get_paths(recursive=True, upn=True):
paths.append(path)
self.assertEqual(len(paths), 3)
self.assertEqual(paths[0].name, "dir1")
self.assertEqual(paths[2].is_directory, False)
@record
def test_create_file_from_file_system_client_async(self):
loop = asyncio.get_event_loop()
loop.run_until_complete(
self._test_create_file_from_file_system_client_async())
class LargeFileTest(StorageTestCase):
def setUp(self):
super(LargeFileTest, self).setUp()
url = self._get_account_url()
self.payload_dropping_policy = PayloadDroppingPolicy()
credential_policy = _format_shared_key_credential(
self.settings.STORAGE_DATA_LAKE_ACCOUNT_NAME,
self.settings.STORAGE_DATA_LAKE_ACCOUNT_KEY)
self.dsc = DataLakeServiceClient(
url,
credential=self.settings.STORAGE_DATA_LAKE_ACCOUNT_KEY,
_additional_pipeline_policies=[
self.payload_dropping_policy, credential_policy
])
loop = asyncio.get_event_loop()
loop.run_until_complete(self.dsc.__aenter__())
self.config = self.dsc._config
self.file_system_name = self.get_resource_name('filesystem')
if not self.is_playback():
file_system = self.dsc.get_file_system_client(
self.file_system_name)
try:
loop = asyncio.get_event_loop()
loop.run_until_complete(
file_system.create_file_system(timeout=5))
except ResourceExistsError:
pass
def tearDown(self):
if not self.is_playback():
try:
loop = asyncio.get_event_loop()
loop.run_until_complete(
self.dsc.delete_file_system(self.file_system_name))
loop.run_until_complete(self.dsc.__aexit__())
except:
pass
return super(LargeFileTest, self).tearDown()
# --Helpers-----------------------------------------------------------------
def _get_directory_reference(self, prefix=TEST_DIRECTORY_PREFIX):
directory_name = self.get_resource_name(prefix)
return directory_name
# --Helpers-----------------------------------------------------------------
async def _test_append_large_stream_without_network(self):
directory_name = self._get_directory_reference()
# Create a directory to put the file under that
directory_client = self.dsc.get_directory_client(
self.file_system_name, directory_name)
await directory_client.create_directory()
file_client = directory_client.get_file_client('filename')
await file_client.create_file()
data = LargeStream(LARGEST_BLOCK_SIZE)
# Act
response = await file_client.append_data(data, 0, LARGEST_BLOCK_SIZE)
self.assertIsNotNone(response)
self.assertEqual(self.payload_dropping_policy.append_counter, 1)
self.assertEqual(self.payload_dropping_policy.append_sizes[0],
LARGEST_BLOCK_SIZE)
@pytest.mark.live_test_only
def test_append_large_stream_without_network(self):
loop = asyncio.get_event_loop()
loop.run_until_complete(
self._test_append_large_stream_without_network())
async def _test_upload_large_stream_without_network(self):
directory_name = self.get_resource_name(TEST_DIRECTORY_PREFIX)
# Create a directory to put the file under that
directory_client = self.dsc.get_directory_client(
self.file_system_name, directory_name)
await directory_client.create_directory()
file_client = directory_client.get_file_client('filename')
await file_client.create_file()
length = 2 * LARGEST_BLOCK_SIZE
data = LargeStream(length)
# Act
response = await file_client.upload_data(data,
length,
overwrite=True,
chunk_size=LARGEST_BLOCK_SIZE)
self.assertIsNotNone(response)
self.assertEqual(self.payload_dropping_policy.append_counter, 2)
self.assertEqual(self.payload_dropping_policy.append_sizes[0],
LARGEST_BLOCK_SIZE)
self.assertEqual(self.payload_dropping_policy.append_sizes[1],
LARGEST_BLOCK_SIZE)
@pytest.mark.live_test_only
def test_upload_large_stream_without_network(self):
loop = asyncio.get_event_loop()
loop.run_until_complete(
self._test_upload_large_stream_without_network())
class DirectoryTest(StorageTestCase):
def setUp(self):
super(DirectoryTest, self).setUp()
url = self._get_account_url()
self.dsc = DataLakeServiceClient(
url,
credential=self.settings.STORAGE_DATA_LAKE_ACCOUNT_KEY,
transport=AiohttpTestTransport())
loop = asyncio.get_event_loop()
loop.run_until_complete(self.dsc.__aenter__())
self.config = self.dsc._config
self.file_system_name = self.get_resource_name('filesystem')
if not self.is_playback():
file_system = self.dsc.get_file_system_client(
self.file_system_name)
try:
loop.run_until_complete(
file_system.create_file_system(timeout=5))
except ResourceExistsError:
pass
def tearDown(self):
if not self.is_playback():
try:
loop = asyncio.get_event_loop()
loop.run_until_complete(
self.dsc.delete_file_system(self.file_system_name))
loop.run_until_complete(self.dsc.__aexit__())
except:
pass
return super(DirectoryTest, self).tearDown()
# --Helpers-----------------------------------------------------------------
def _get_directory_reference(self, prefix=TEST_DIRECTORY_PREFIX):
directory_name = self.get_resource_name(prefix)
return directory_name
async def _create_directory_and_get_directory_client(
self, directory_name=None):
directory_name = directory_name if directory_name else self._get_directory_reference(
)
directory_client = self.dsc.get_directory_client(
self.file_system_name, directory_name)
await directory_client.create_directory()
return directory_client
async def _create_file_system(self):
return await self.dsc.create_file_system(
self._get_file_system_reference())
# --Helpers-----------------------------------------------------------------
async def _test_create_directory(self):
# Arrange
directory_name = self._get_directory_reference()
content_settings = ContentSettings(content_language='spanish',
content_disposition='inline')
# Act
directory_client = self.dsc.get_directory_client(
self.file_system_name, directory_name)
created = await directory_client.create_directory(
content_settings=content_settings)
# Assert
self.assertTrue(created)
@record
def test_create_directory_async(self):
loop = asyncio.get_event_loop()
loop.run_until_complete(self._test_create_directory())
async def _test_using_oauth_token_credential_to_create_directory(self):
# generate a token with directory level create permission
directory_name = self._get_directory_reference()
token_credential = self.generate_async_oauth_token()
directory_client = DataLakeDirectoryClient(self.dsc.url,
self.file_system_name,
directory_name,
credential=token_credential)
response = await directory_client.create_directory()
self.assertIsNotNone(response)
@record
def test_using_oauth_token_credential_to_create_directory_async(self):
loop = asyncio.get_event_loop()
loop.run_until_complete(
self._test_using_oauth_token_credential_to_create_directory())
async def _test_create_directory_with_match_conditions(self):
# Arrange
directory_name = self._get_directory_reference()
# Act
directory_client = self.dsc.get_directory_client(
self.file_system_name, directory_name)
created = await directory_client.create_directory(
match_condition=MatchConditions.IfMissing)
# Assert
self.assertTrue(created)
@record
def test_create_directory_with_match_conditions_async(self):
loop = asyncio.get_event_loop()
loop.run_until_complete(
self._test_create_directory_with_match_conditions())
async def _test_create_directory_with_permission(self):
# Arrange
directory_name = self._get_directory_reference()
# Act
directory_client = self.dsc.get_directory_client(
self.file_system_name, directory_name)
created = await directory_client.create_directory(
permissions="rwxr--r--", umask="0000")
prop = await directory_client.get_access_control()
# Assert
self.assertTrue(created)
self.assertEqual(prop['permissions'], 'rwxr--r--')
@record
def test_create_directory_with_permission_async(self):
loop = asyncio.get_event_loop()
loop.run_until_complete(self._test_create_directory_with_permission())
async def _test_create_directory_with_content_settings(self):
# Arrange
directory_name = self._get_directory_reference()
content_settings = ContentSettings(content_language='spanish',
content_disposition='inline')
# Act
directory_client = self.dsc.get_directory_client(
self.file_system_name, directory_name)
created = await directory_client.create_directory(
content_settings=content_settings)
# Assert
self.assertTrue(created)
@record
def test_create_directory_with_content_settings_async(self):
loop = asyncio.get_event_loop()
loop.run_until_complete(
self._test_create_directory_with_content_settings())
async def _test_create_directory_with_metadata(self):
# Arrange
directory_name = self._get_directory_reference()
metadata = {'hello': 'world', 'number': '42'}
# Act
directory_client = self.dsc.get_directory_client(
self.file_system_name, directory_name)
created = await directory_client.create_directory(metadata=metadata)
properties = await directory_client.get_directory_properties()
# Assert
self.assertTrue(created)
@record
def test_create_directory_with_metadata_async(self):
loop = asyncio.get_event_loop()
loop.run_until_complete(self._test_create_directory_with_metadata())
async def _test_delete_directory(self):
# Arrange
directory_name = self._get_directory_reference()
metadata = {'hello': 'world', 'number': '42'}
directory_client = self.dsc.get_directory_client(
self.file_system_name, directory_name)
await directory_client.create_directory(metadata=metadata)
response = await directory_client.delete_directory()
# Assert
self.assertIsNone(response)
@record
def test_delete_directory_async(self):
loop = asyncio.get_event_loop()
loop.run_until_complete(self._test_delete_directory())
async def _test_delete_directory_with_if_modified_since(self):
# Arrange
directory_name = self._get_directory_reference()
directory_client = self.dsc.get_directory_client(
self.file_system_name, directory_name)
await directory_client.create_directory()
prop = await directory_client.get_directory_properties()
with self.assertRaises(ResourceModifiedError):
await directory_client.delete_directory(
if_modified_since=prop['last_modified'])
@record
def test_delete_directory_with_if_modified_since_async(self):
loop = asyncio.get_event_loop()
loop.run_until_complete(
self._test_delete_directory_with_if_modified_since())
async def _test_create_sub_directory_and_delete_sub_directory(self):
# Arrange
directory_name = self._get_directory_reference()
metadata = {'hello': 'world', 'number': '42'}
# Create a directory first, to prepare for creating sub directory
directory_client = self.dsc.get_directory_client(
self.file_system_name, directory_name)
await directory_client.create_directory(metadata=metadata)
# Create sub directory from the current directory
sub_directory_name = 'subdir'
sub_directory_created = await directory_client.create_sub_directory(
sub_directory_name)
# to make sure the sub directory was indeed created by get sub_directory properties from sub directory client
sub_directory_client = self.dsc.get_directory_client(
self.file_system_name, directory_name + '/' + sub_directory_name)
sub_properties = await sub_directory_client.get_directory_properties()
# Assert
self.assertTrue(sub_directory_created)
self.assertTrue(sub_properties)
# Act
await directory_client.delete_sub_directory(sub_directory_name)
with self.assertRaises(ResourceNotFoundError):
await sub_directory_client.get_directory_properties()
@record
def test_create_sub_directory_and_delete_sub_directory_async(self):
loop = asyncio.get_event_loop()
loop.run_until_complete(
self._test_create_sub_directory_and_delete_sub_directory())
async def _test_set_access_control(self):
directory_name = self._get_directory_reference()
metadata = {'hello': 'world', 'number': '42'}
directory_client = self.dsc.get_directory_client(
self.file_system_name, directory_name)
await directory_client.create_directory(metadata=metadata)
response = await directory_client.set_access_control(permissions='0777'
)
# Assert
self.assertIsNotNone(response)
@record
def test_set_access_control_async(self):
loop = asyncio.get_event_loop()
loop.run_until_complete(self._test_set_access_control())
async def _test_set_access_control_with_acl(self):
directory_name = self._get_directory_reference()
metadata = {'hello': 'world', 'number': '42'}
directory_client = self.dsc.get_directory_client(
self.file_system_name, directory_name)
await directory_client.create_directory(metadata=metadata)
acl = 'user::rwx,group::r-x,other::rwx'
await directory_client.set_access_control(acl=acl)
access_control = await directory_client.get_access_control()
# Assert
self.assertIsNotNone(access_control)
self.assertEqual(acl, access_control['acl'])
@record
def test_set_access_control_with_acl_async(self):
loop = asyncio.get_event_loop()
loop.run_until_complete(self._test_set_access_control_with_acl())
async def _test_set_access_control_if_none_modified(self):
directory_name = self._get_directory_reference()
directory_client = self.dsc.get_directory_client(
self.file_system_name, directory_name)
resp = await directory_client.create_directory()
response = await directory_client.set_access_control(
permissions='0777',
etag=resp['etag'],
match_condition=MatchConditions.IfNotModified)
# Assert
self.assertIsNotNone(response)
@record
def test_set_access_control_if_none_modified_async(self):
loop = asyncio.get_event_loop()
loop.run_until_complete(
self._test_set_access_control_if_none_modified())
async def _test_get_access_control(self):
directory_name = self._get_directory_reference()
metadata = {'hello': 'world', 'number': '42'}
directory_client = self.dsc.get_directory_client(
self.file_system_name, directory_name)
await directory_client.create_directory(metadata=metadata,
permissions='0777')
# Act
response = await directory_client.get_access_control()
# Assert
self.assertIsNotNone(response)
@record
def test_get_access_control_async(self):
loop = asyncio.get_event_loop()
loop.run_until_complete(self._test_get_access_control())
async def _test_get_access_control_with_match_conditions(self):
directory_name = self._get_directory_reference()
directory_client = self.dsc.get_directory_client(
self.file_system_name, directory_name)
resp = await directory_client.create_directory(permissions='0777',
umask='0000')
# Act
response = await directory_client.get_access_control(
etag=resp['etag'], match_condition=MatchConditions.IfNotModified)
# Assert
self.assertIsNotNone(response)
self.assertEquals(response['permissions'], 'rwxrwxrwx')
@record
def test_get_access_control_with_match_conditions_async(self):
loop = asyncio.get_event_loop()
loop.run_until_complete(
self._test_get_access_control_with_match_conditions())
async def _test_rename_from(self):
content_settings = ContentSettings(content_language='spanish',
content_disposition='inline')
directory_name = self._get_directory_reference()
directory_client = self.dsc.get_directory_client(
self.file_system_name, directory_name)
await directory_client.create_directory()
new_name = "newname"
new_directory_client = self.dsc.get_directory_client(
self.file_system_name, new_name)
await new_directory_client._rename_path(
'/' + self.file_system_name + '/' + directory_name,
content_settings=content_settings)
properties = await new_directory_client.get_directory_properties()
self.assertIsNotNone(properties)
self.assertIsNone(properties.get('content_settings'))
@record
def test_rename_from_async(self):
loop = asyncio.get_event_loop()
loop.run_until_complete(self._test_rename_from())
async def _test_rename_from_a_shorter_directory_to_longer_directory(self):
# TODO: investigate why rename shorter path to a longer one does not work
pytest.skip("")
directory_name = self._get_directory_reference()
await self._create_directory_and_get_directory_client(
directory_name="old")
new_name = "newname"
new_directory_client = await self._create_directory_and_get_directory_client(
directory_name=new_name)
new_directory_client = await new_directory_client.create_sub_directory(
"newsub")
await new_directory_client._rename_path('/' + self.file_system_name +
'/' + directory_name)
properties = await new_directory_client.get_directory_properties()
self.assertIsNotNone(properties)
@record
def test_rename_from_a_shorter_directory_to_longer_directory_async(self):
loop = asyncio.get_event_loop()
loop.run_until_complete(
self._test_rename_from_a_shorter_directory_to_longer_directory())
async def _test_rename_from_a_directory_in_another_file_system(self):
# create a file dir1 under file system1
old_file_system_name = self._get_directory_reference("oldfilesystem")
old_dir_name = "olddir"
old_client = self.dsc.get_file_system_client(old_file_system_name)
time.sleep(30)
await old_client.create_file_system()
await old_client.create_directory(old_dir_name)
# create a dir2 under file system2
new_name = "newname"
time.sleep(5)
new_directory_client = await self._create_directory_and_get_directory_client(
directory_name=new_name)
new_directory_client = await new_directory_client.create_sub_directory(
"newsub")
# rename dir1 under file system1 to dir2 under file system2
await new_directory_client._rename_path('/' + old_file_system_name +
'/' + old_dir_name)
properties = await new_directory_client.get_directory_properties()
self.assertIsNotNone(properties)
@record
def test_rename_from_a_directory_in_another_file_system_async(self):
loop = asyncio.get_event_loop()
loop.run_until_complete(
self._test_rename_from_a_directory_in_another_file_system())
async def _test_rename_to_an_existing_directory_in_another_file_system(
self):
# create a file dir1 under file system1
destination_file_system_name = self._get_directory_reference(
"destfilesystem")
destination_dir_name = "destdir"
fs_client = self.dsc.get_file_system_client(
destination_file_system_name)
time.sleep(30)
await fs_client.create_file_system()
destination_directory_client = await fs_client.create_directory(
destination_dir_name)
# create a dir2 under file system2
source_name = "source"
source_directory_client = await self._create_directory_and_get_directory_client(
directory_name=source_name)
source_directory_client = await source_directory_client.create_sub_directory(
"subdir")
# rename dir2 under file system2 to dir1 under file system1
res = await source_directory_client.rename_directory(
'/' + destination_file_system_name + '/' + destination_dir_name)
# the source directory has been renamed to destination directory, so it cannot be found
with self.assertRaises(HttpResponseError):
await source_directory_client.get_directory_properties()
self.assertEquals(res.url, destination_directory_client.url)
@record
def test_rename_to_an_existing_directory_in_another_file_system_async(
self):
loop = asyncio.get_event_loop()
loop.run_until_complete(
self._test_rename_to_an_existing_directory_in_another_file_system(
))
async def _test_rename_with_none_existing_destination_condition_and_source_unmodified_condition(
self):
non_existing_dir_name = "nonexistingdir"
# create a file system1
destination_file_system_name = self._get_directory_reference(
"destfilesystem")
fs_client = self.dsc.get_file_system_client(
destination_file_system_name)
await fs_client.create_file_system()
# create a dir2 under file system2
source_name = "source"
source_directory_client = await self._create_directory_and_get_directory_client(
directory_name=source_name)
source_directory_client = await source_directory_client.create_sub_directory(
"subdir")
# rename dir2 under file system2 to a non existing directory under file system1,
# when dir1 does not exist and dir2 wasn't modified
properties = await source_directory_client.get_directory_properties()
etag = properties['etag']
res = await source_directory_client.rename_directory(
'/' + destination_file_system_name + '/' + non_existing_dir_name,
match_condition=MatchConditions.IfMissing,
source_etag=etag,
source_match_condition=MatchConditions.IfNotModified)
# the source directory has been renamed to destination directory, so it cannot be found
with self.assertRaises(HttpResponseError):
await source_directory_client.get_directory_properties()
self.assertEquals(non_existing_dir_name, res.path_name)
@record
def test_rename_with_none_existing_destination_condition_and_source_unmodified_condition_async(
self):
loop = asyncio.get_event_loop()
loop.run_until_complete(
self.
_test_rename_with_none_existing_destination_condition_and_source_unmodified_condition(
))
async def _test_rename_to_an_non_existing_directory_in_another_file_system(
self):
# create a file dir1 under file system1
destination_file_system_name = self._get_directory_reference(
"destfilesystem")
non_existing_dir_name = "nonexistingdir"
fs_client = self.dsc.get_file_system_client(
destination_file_system_name)
await fs_client.create_file_system()
# create a dir2 under file system2
source_name = "source"
source_directory_client = await self._create_directory_and_get_directory_client(
directory_name=source_name)
source_directory_client = await source_directory_client.create_sub_directory(
"subdir")
# rename dir2 under file system2 to dir1 under file system1
res = await source_directory_client.rename_directory(
'/' + destination_file_system_name + '/' + non_existing_dir_name)
# the source directory has been renamed to destination directory, so it cannot be found
with self.assertRaises(HttpResponseError):
await source_directory_client.get_directory_properties()
self.assertEquals(non_existing_dir_name, res.path_name)
@record
def test_rename_to_an_non_existing_directory_in_another_file_system_async(
self):
loop = asyncio.get_event_loop()
loop.run_until_complete(
self.
_test_rename_to_an_non_existing_directory_in_another_file_system())
async def _test_rename_directory_to_non_empty_directory(self):
# TODO: investigate why rename non empty dir doesn't work
pytest.skip("")
dir1 = await self._create_directory_and_get_directory_client("dir1")
await dir1.create_sub_directory("subdir")
dir2 = await self._create_directory_and_get_directory_client("dir2")
await dir2.rename_directory(dir1.file_system_name + '/' +
dir1.path_name)
with self.assertRaises(HttpResponseError):
await dir2.get_directory_properties()
@record
def test_rename_directory_to_non_empty_directory_async(self):
loop = asyncio.get_event_loop()
loop.run_until_complete(
self._test_rename_directory_to_non_empty_directory())
async def _test_get_properties(self):
# Arrange
directory_name = self._get_directory_reference()
metadata = {'hello': 'world', 'number': '42'}
directory_client = self.dsc.get_directory_client(
self.file_system_name, directory_name)
await directory_client.create_directory(metadata=metadata)
properties = await directory_client.get_directory_properties()
# Assert
self.assertTrue(properties)
self.assertIsNotNone(properties.metadata)
self.assertEqual(properties.metadata['hello'], metadata['hello'])
@record
def test_get_properties_async(self):
loop = asyncio.get_event_loop()
loop.run_until_complete(self._test_get_properties())
async def _test_using_directory_sas_to_read(self):
# SAS URL is calculated from storage key, so this test runs live only
if TestMode.need_recording_file(self.test_mode):
return
client = await self._create_directory_and_get_directory_client()
directory_name = client.path_name
# generate a token with directory level read permission
token = generate_directory_sas(
self.dsc.account_name,
self.file_system_name,
directory_name,
self.dsc.credential.account_key,
permission=DirectorySasPermissions(read=True),
expiry=datetime.utcnow() + timedelta(hours=1),
)
directory_client = DataLakeDirectoryClient(self.dsc.url,
self.file_system_name,
directory_name,
credential=token)
access_control = await directory_client.get_access_control()
self.assertIsNotNone(access_control)
@record
def test_using_directory_sas_to_read_async(self):
loop = asyncio.get_event_loop()
loop.run_until_complete(self._test_using_directory_sas_to_read())
async def _test_using_directory_sas_to_create(self):
# SAS URL is calculated from storage key, so this test runs live only
if TestMode.need_recording_file(self.test_mode):
return
# generate a token with directory level create permission
directory_name = self._get_directory_reference()
token = generate_directory_sas(
self.dsc.account_name,
self.file_system_name,
directory_name,
self.dsc.credential.account_key,
permission=DirectorySasPermissions(create=True),
expiry=datetime.utcnow() + timedelta(hours=1),
)
directory_client = DataLakeDirectoryClient(self.dsc.url,
self.file_system_name,
directory_name,
credential=token)
response = await directory_client.create_directory()
self.assertIsNotNone(response)
@record
def test_using_directory_sas_to_create_async(self):
loop = asyncio.get_event_loop()
loop.run_until_complete(self._test_using_directory_sas_to_create())
async def data_lake_service_sample():
# Instantiate a DataLakeServiceClient using a connection string
# [START create_datalake_service_client]
from azure.storage.filedatalake.aio import DataLakeServiceClient
datalake_service_client = DataLakeServiceClient.from_connection_string(
connection_string)
# [END create_datalake_service_client]
# Instantiate a DataLakeServiceClient Azure Identity credentials.
# [START create_datalake_service_client_oauth]
from azure.identity.aio import ClientSecretCredential
token_credential = ClientSecretCredential(
active_directory_tenant_id,
active_directory_application_id,
active_directory_application_secret,
)
datalake_service_client = DataLakeServiceClient(
"https://{}.dfs.core.windows.net".format(account_name),
credential=token_credential)
# [END create_datalake_service_client_oauth]
async with datalake_service_client:
# get user delegation key
# [START get_user_delegation_key]
from datetime import datetime, timedelta
user_delegation_key = await datalake_service_client.get_user_delegation_key(
datetime.utcnow(),
datetime.utcnow() + timedelta(hours=1))
# [END get_user_delegation_key]
# Create file systems
# [START create_file_system_from_service_client]
await datalake_service_client.create_file_system("filesystem")
# [END create_file_system_from_service_client]
file_system_client = await datalake_service_client.create_file_system(
"anotherfilesystem")
# List file systems
# [START list_file_systems]
file_systems = datalake_service_client.list_file_systems()
async for file_system in file_systems:
print(file_system.name)
# [END list_file_systems]
# Get Clients from DataLakeServiceClient
file_system_client = datalake_service_client.get_file_system_client(
file_system_client.file_system_name)
# [START get_directory_client_from_service_client]
directory_client = datalake_service_client.get_directory_client(
file_system_client.file_system_name, "mydirectory")
# [END get_directory_client_from_service_client]
# [START get_file_client_from_service_client]
file_client = datalake_service_client.get_file_client(
file_system_client.file_system_name, "myfile")
# [END get_file_client_from_service_client]
# Create file and set properties
metadata = {'hello': 'world', 'number': '42'}
from azure.storage.filedatalake import ContentSettings
content_settings = ContentSettings(content_language='spanish',
content_disposition='inline')
await file_client.create_file(content_settings=content_settings)
await file_client.set_metadata(metadata=metadata)
file_props = await file_client.get_file_properties()
print(file_props.metadata)
# Create file/directory and set properties
await directory_client.create_directory(
content_settings=content_settings, metadata=metadata)
dir_props = await directory_client.get_directory_properties()
print(dir_props.metadata)
# Delete File Systems
# [START delete_file_system_from_service_client]
await datalake_service_client.delete_file_system("filesystem")
# [END delete_file_system_from_service_client]
await file_system_client.delete_file_system()
await token_credential.close()
