def test_list_non_empty(self, client):
base_path = "/path/"
# Create some files to return
dir_prefix = BlobPrefix()
dir_prefix.name = base_path + "dir"
blob1_props = BlobProperties()
blob1_props.size = 42
blob1_props.name = base_path + "file1"
blob2_props = BlobProperties()
blob2_props.size = 0
blob2_props.name = base_path + "file2"
client.get_container_client().walk_blobs.return_value = MockBlobList(
[dir_prefix, blob1_props, blob2_props]
)
store = AzureBlobStoreService()
store.set_connection(connection=client)
key_path = self.wasbs_base + base_path
results = store.list(key=key_path)
assert len(results["blobs"]) == 2
assert len(results["prefixes"]) == 1
assert results["prefixes"][0] == "dir"
assert results["blobs"][0][0] == "file1"
assert results["blobs"][0][1] == 42
assert results["blobs"][1][0] == "file2"
assert results["blobs"][1][1] == 0
def test_download_dir_with_basename(self, client):
dirname1 = tempfile.mkdtemp()
dirname2 = tempfile.mkdtemp(prefix=dirname1 + "/")
def mkfile(fname):
return open(fname, "w")
client.return_value.get_blob_to_path.side_effect = mkfile
store = AzureBlobStoreService()
store.set_connection(connection=client)
blob_path = "/path/to/"
azure_url = self.wasbs_base + blob_path
rel_path2 = dirname2.split("/")[-1]
# Mock return list
obj_mock1 = BlobProperties()
obj_mock1.size = 42
obj_mock1.name = blob_path + "foo/test1.txt"
obj_mock2 = BlobProperties()
obj_mock2.size = 42
obj_mock2.name = blob_path + "foo/test2.txt"
subdir_mock = BlobPrefix()
subdir_mock.name = blob_path + "foo/" + rel_path2
obj_mock3 = BlobProperties()
obj_mock3.size = 42
obj_mock3.name = blob_path + "foo/" + rel_path2 + "/" + "test3.txt"
# Create some files to return
def list_side_effect(container_name, prefix, delimiter="/", marker=None):
if prefix == blob_path + "foo/":
return MockBlobList([subdir_mock, obj_mock1, obj_mock2])
return MockBlobList([obj_mock3])
client.return_value.walk_blobs.side_effect = list_side_effect
dirname3 = tempfile.mkdtemp()
# Test without basename
store.download_dir(
blob=azure_url + "foo", local_path=dirname3, use_basename=True
)
call_args_list = client.get_container_client().download_blob.call_args_list
for call_args in call_args_list:
call_arg1 = call_args[0]
if call_arg1 == "{}foo/test1.txt".format(blob_path):
assert os.path.exists("{}/foo/test1.txt".format(dirname3))
elif call_arg1 == "{}foo/test2.txt".format(blob_path):
assert os.path.exists("{}/foo/test2.txt".format(dirname3))
elif call_arg1 == "{}foo/{}/test3.txt".format(blob_path, rel_path2):
assert os.path.exists("{}/foo/{}/test3.txt".format(dirname3, rel_path2))
else:
assert False
def test_download_directory_artifact_succeeds_when_artifact_root_is_blob_container_root(
mock_client, tmpdir):
repo = AzureBlobArtifactRepository(TEST_BLOB_CONTAINER_ROOT, mock_client)
subdir_path = "my_directory"
dir_prefix = BlobPrefix()
dir_prefix.name = subdir_path
file_path_1 = "file_1"
file_path_2 = "file_2"
blob_props_1 = BlobProperties()
blob_props_1.size = 42
blob_props_1.name = posixpath.join(subdir_path, file_path_1)
blob_props_2 = BlobProperties()
blob_props_2.size = 42
blob_props_2.name = posixpath.join(subdir_path, file_path_2)
def get_mock_listing(*args, **kwargs):
"""
Produces a mock listing that only contains content if the specified prefix is the artifact
root or a relevant subdirectory. This allows us to mock `list_artifacts` during the
`_download_artifacts_into` subroutine without recursively listing the same artifacts at
every level of the directory traversal.
"""
# pylint: disable=unused-argument
if posixpath.abspath(kwargs["name_starts_with"]) == "/":
return MockBlobList([dir_prefix])
if posixpath.abspath(
kwargs["name_starts_with"]) == posixpath.abspath(subdir_path):
return MockBlobList([blob_props_1, blob_props_2])
else:
return MockBlobList([])
def create_file(buffer):
fname = os.path.basename(buffer.name)
f = tmpdir.join(fname)
f.write("hello world!")
mock_client.get_container_client(
).walk_blobs.side_effect = get_mock_listing
mock_client.get_container_client().download_blob(
).readinto.side_effect = create_file
# Ensure that the root directory can be downloaded successfully
repo.download_artifacts("")
# Ensure that the `mkfile` side effect copied all of the download artifacts into `tmpdir`
dir_contents = os.listdir(tmpdir.strpath)
assert file_path_1 in dir_contents
assert file_path_2 in dir_contents
def test_download_artifact_throws_value_error_when_listed_blobs_do_not_contain_artifact_root_prefix(
mock_client, ):
repo = AzureBlobArtifactRepository(TEST_URI, mock_client)
# Create a "bad blob" with a name that is not prefixed by the root path of the artifact store
bad_blob_props = BlobProperties()
bad_blob_props.size = 42
bad_blob_props.name = "file_path"
def get_mock_listing(*args, **kwargs):
"""
Produces a mock listing that only contains content if the
specified prefix is the artifact root. This allows us to mock
`list_artifacts` during the `_download_artifacts_into` subroutine
without recursively listing the same artifacts at every level of the
directory traversal.
"""
# pylint: disable=unused-argument
if posixpath.abspath(kwargs["name_starts_with"]) == posixpath.abspath(
TEST_ROOT_PATH):
# Return a blob that is not prefixed by the root path of the artifact store. This
# should result in an exception being raised
return MockBlobList([bad_blob_props])
else:
return MockBlobList([])
mock_client.get_container_client(
).walk_blobs.side_effect = get_mock_listing
with pytest.raises(
MlflowException,
match="Azure blob does not begin with the specified artifact path"
):
repo.download_artifacts("")
def test_list_artifacts(mock_client, root_path):
repo = AzureBlobArtifactRepository(
posixpath.join(TEST_BLOB_CONTAINER_ROOT, root_path), mock_client
)
# Create some files to return
dir_prefix = BlobPrefix()
dir_prefix.name = posixpath.join(TEST_ROOT_PATH, "dir")
blob_props = BlobProperties()
blob_props.size = 42
blob_props.name = posixpath.join(TEST_ROOT_PATH, "file")
mock_client.get_container_client().walk_blobs.return_value = MockBlobList(
[dir_prefix, blob_props]
)
artifacts = repo.list_artifacts()
mock_client.get_container_client().walk_blobs.assert_called_with(name_starts_with="some/path/")
assert artifacts[0].path == "dir"
assert artifacts[0].is_dir is True
assert artifacts[0].file_size is None
assert artifacts[1].path == "file"
assert artifacts[1].is_dir is False
assert artifacts[1].file_size == 42
def test_list_artifacts_single_file(mock_client):
repo = AzureBlobArtifactRepository(TEST_URI, mock_client)
# Evaluate single file
blob_props = BlobProperties()
blob_props.name = posixpath.join(TEST_ROOT_PATH, "file")
mock_client.get_container_client().walk_blobs.return_value = MockBlobList([blob_props])
assert repo.list_artifacts("file") == []
def get_blob_properties(filename):
logger = logging.getLogger(__name__)
props = BlobProperties()
content = ContentSettings()
logger.debug('Settings props for file %s', filename)
props.name = filename
if '.json' in filename:
content.content_type = 'application/json'
elif '.html' in filename:
content.content_type = 'text/html'
else:
content.content_type = 'text/plain'
logger.debug('Content-type set to %s', content.content_type)
content.content_disposition = f'inline; filename={filename}'
props.content_settings = content
return props