def __init__(self, options):
"""Initializes a Dataflow API client object."""
self.standard_options = options.view_as(StandardOptions)
self.google_cloud_options = options.view_as(GoogleCloudOptions)
if _use_fnapi(options):
self.environment_version = _FNAPI_ENVIRONMENT_MAJOR_VERSION
else:
self.environment_version = _LEGACY_ENVIRONMENT_MAJOR_VERSION
if self.google_cloud_options.no_auth:
credentials = None
else:
credentials = get_service_credentials()
# Use 60 second socket timeout avoid hangs during network flakiness.
http_client = httplib2.Http(timeout=60)
self._client = dataflow.DataflowV1b3(
url=self.google_cloud_options.dataflow_endpoint,
credentials=credentials,
get_credentials=(not self.google_cloud_options.no_auth),
http=http_client)
self._storage_client = storage.StorageV1(
url='https://www.googleapis.com/storage/v1',
credentials=credentials,
get_credentials=(not self.google_cloud_options.no_auth),
http=http_client)
def __init__(self, options):
"""Initializes a Dataflow API client object."""
self.standard_options = options.view_as(StandardOptions)
self.google_cloud_options = options.view_as(GoogleCloudOptions)
if _use_fnapi(options):
self.environment_version = _FNAPI_ENVIRONMENT_MAJOR_VERSION
else:
self.environment_version = _LEGACY_ENVIRONMENT_MAJOR_VERSION
if self.google_cloud_options.no_auth:
credentials = None
else:
credentials = get_service_credentials()
http_client = get_new_http()
self._client = dataflow.DataflowV1b3(
url=self.google_cloud_options.dataflow_endpoint,
credentials=credentials,
get_credentials=(not self.google_cloud_options.no_auth),
http=http_client,
response_encoding=get_response_encoding())
self._storage_client = storage.StorageV1(
url='https://www.googleapis.com/storage/v1',
credentials=credentials,
get_credentials=(not self.google_cloud_options.no_auth),
http=http_client,
response_encoding=get_response_encoding())
def assert_bucket_exists(bucket_name):
# type: (str) -> None
"""Asserts whether the specified GCS bucket with the name
bucket_name exists.
Logs an error and raises a ValueError if the bucket does not exist.
Logs a warning if the bucket cannot be verified to exist.
"""
try:
from apitools.base.py.exceptions import HttpError
storage_client = storage.StorageV1(
credentials=auth.get_service_credentials(),
get_credentials=False,
http=get_new_http(),
response_encoding='utf8')
request = storage.StorageBucketsGetRequest(bucket=bucket_name)
storage_client.buckets.Get(request)
except HttpError as e:
if e.status_code == 404:
_LOGGER.error('%s bucket does not exist!', bucket_name)
raise ValueError('Invalid GCS bucket provided!')
else:
_LOGGER.warning(
'HttpError - unable to verify whether bucket %s exists',
bucket_name)
except ImportError:
_LOGGER.warning(
'ImportError - unable to verify whether bucket %s exists',
bucket_name)
def __init__(self, storage_client=None):
if storage_client is None:
storage_client = storage.StorageV1(
credentials=auth.get_service_credentials(),
get_credentials=False,
http=get_new_http(),
response_encoding='utf8')
self.client = storage_client
self._rewrite_cb = None
def __init__(self, storage_client=None, pipeline_options=None):
if storage_client is None:
storage_client = storage.StorageV1(
credentials=auth.get_service_credentials(pipeline_options),
get_credentials=False,
http=get_new_http(),
response_encoding='utf8',
additional_http_headers={
"User-Agent": "apache-beam-%s" % apache_beam.__version__
})
self.client = storage_client
self._rewrite_cb = None
self.bucket_to_project_number = {}
def __new__(cls, storage_client=None):
if storage_client:
return super(GcsIO, cls).__new__(cls, storage_client)
else:
# Create a single storage client for each thread. We would like to avoid
# creating more than one storage client for each thread, since each
# initialization requires the relatively expensive step of initializing
# credentaials.
local_state = threading.local()
if getattr(local_state, 'gcsio_instance', None) is None:
credentials = auth.get_service_credentials()
storage_client = storage.StorageV1(credentials=credentials)
local_state.gcsio_instance = (super(GcsIO, cls).__new__(
cls, storage_client))
local_state.gcsio_instance.client = storage_client
return local_state.gcsio_instance
def __init__(self, options):
super().__init__(options)
self._google_cloud_options = options.view_as(GoogleCloudOptions)
if self._google_cloud_options.no_auth:
credentials = None
else:
credentials = get_service_credentials()
self._storage_client = storage.StorageV1(
url='https://www.googleapis.com/storage/v1',
credentials=credentials,
get_credentials=(not self._google_cloud_options.no_auth),
http=get_new_http(),
response_encoding='utf8')
if not self._docker_registry_push_url:
self._docker_registry_push_url = (
'gcr.io/%s' % self._google_cloud_options.project)
def __init__(self, options, environment_version):
"""Initializes a Dataflow API client object."""
self.standard_options = options.view_as(StandardOptions)
self.google_cloud_options = options.view_as(GoogleCloudOptions)
self.environment_version = environment_version
if self.google_cloud_options.no_auth:
credentials = None
else:
credentials = get_service_credentials()
self._client = dataflow.DataflowV1b3(
url=self.google_cloud_options.dataflow_endpoint,
credentials=credentials,
get_credentials=(not self.google_cloud_options.no_auth))
self._storage_client = storage.StorageV1(
url='https://www.googleapis.com/storage/v1',
credentials=credentials,
get_credentials=(not self.google_cloud_options.no_auth))
def __new__(cls, storage_client=None):
if storage_client:
# This path is only used for testing.
return super(GcsIO, cls).__new__(cls)
else:
# Create a single storage client for each thread. We would like to avoid
# creating more than one storage client for each thread, since each
# initialization requires the relatively expensive step of initializing
# credentaials.
local_state = threading.local()
if getattr(local_state, 'gcsio_instance', None) is None:
credentials = auth.get_service_credentials()
storage_client = storage.StorageV1(
credentials=credentials,
http=httplib2.Http(timeout=DEFAULT_HTTP_TIMEOUT_SECONDS))
local_state.gcsio_instance = (super(GcsIO, cls).__new__(
cls, storage_client))
local_state.gcsio_instance.client = storage_client
return local_state.gcsio_instance
def __new__(cls, storage_client=None):
if storage_client:
# This path is only used for testing.
return super(GcsIO, cls).__new__(cls)
else:
# Create a single storage client for each thread. We would like to avoid
# creating more than one storage client for each thread, since each
# initialization requires the relatively expensive step of initializing
# credentaials.
local_state = threading.local()
if getattr(local_state, 'gcsio_instance', None) is None:
credentials = auth.get_service_credentials()
storage_client = storage.StorageV1(
credentials=credentials,
get_credentials=False,
http=get_new_http(),
response_encoding=None if sys.version_info[0] < 3 else 'utf8')
local_state.gcsio_instance = super(GcsIO, cls).__new__(cls)
local_state.gcsio_instance.client = storage_client
return local_state.gcsio_instance
def __init__(self, options):
super().__init__(options)
self._google_cloud_options = options.view_as(GoogleCloudOptions)
self._cloud_build_machine_type = self._get_cloud_build_machine_type_enum(
options.view_as(SetupOptions).cloud_build_machine_type)
if self._google_cloud_options.no_auth:
credentials = None
else:
credentials = get_service_credentials(options)
self._storage_client = storage.StorageV1(
url='https://www.googleapis.com/storage/v1',
credentials=credentials,
get_credentials=(not self._google_cloud_options.no_auth),
http=get_new_http(),
response_encoding='utf8')
self._cloudbuild_client = cloudbuild.CloudbuildV1(
credentials=credentials,
get_credentials=(not self._google_cloud_options.no_auth),
http=get_new_http(),
response_encoding='utf8')
if not self._docker_registry_push_url:
self._docker_registry_push_url = (
'gcr.io/%s/prebuilt_beam_sdk' % self._google_cloud_options.project)