def __init__(self, client, path, buffer_size):
self._client = client
self._path = path
self._bucket, self._name = parse_gcs_path(path)
self._buffer_size = buffer_size
# Get object state.
self._get_request = (storage.StorageObjectsGetRequest(
bucket=self._bucket, object=self._name))
try:
metadata = self._get_object_metadata(self._get_request)
except HttpError as http_error:
if http_error.status_code == 404:
raise IOError(errno.ENOENT, 'Not found: %s' % self._path)
else:
logging.error('HTTP error while requesting file %s: %s',
self._path, http_error)
raise
self._size = metadata.size
# Ensure read is from file of the correct generation.
self._get_request.generation = metadata.generation
# Initialize read buffer state.
self._download_stream = io.BytesIO()
self._downloader = transfer.Download(self._download_stream,
auto_transfer=False,
chunksize=self._buffer_size)
self._client.objects.Get(self._get_request, download=self._downloader)
def __init__(self, client, path, buffer_size=DEFAULT_READ_BUFFER_SIZE):
self.client = client
self.path = path
self.bucket, self.name = parse_gcs_path(path)
self.buffer_size = buffer_size
# Get object state.
get_request = (storage.StorageObjectsGetRequest(bucket=self.bucket,
object=self.name))
try:
metadata = self._get_object_metadata(get_request)
except HttpError as http_error:
if http_error.status_code == 404:
raise IOError(errno.ENOENT, 'Not found: %s' % self.path)
else:
logging.error('HTTP error while requesting file %s: %s',
self.path, http_error)
raise
self.size = metadata.size
# Ensure read is from file of the correct generation.
get_request.generation = metadata.generation
# Initialize read buffer state.
self.download_stream = StringIO.StringIO()
self.downloader = transfer.Download(self.download_stream,
auto_transfer=False)
self.client.objects.Get(get_request, download=self.downloader)
self.position = 0
self.buffer = ''
self.buffer_start_position = 0
self.closed = False
def __init__(self, client, path, buffer_size, get_project_number):
self._client = client
self._path = path
self._bucket, self._name = parse_gcs_path(path)
self._buffer_size = buffer_size
self._get_project_number = get_project_number
project_number = self._get_project_number(self._bucket)
# Create a request count metric
resource = resource_identifiers.GoogleCloudStorageBucket(self._bucket)
labels = {
monitoring_infos.SERVICE_LABEL: 'Storage',
monitoring_infos.METHOD_LABEL: 'Objects.get',
monitoring_infos.RESOURCE_LABEL: resource,
monitoring_infos.GCS_BUCKET_LABEL: self._bucket,
monitoring_infos.GCS_PROJECT_ID_LABEL: project_number
}
service_call_metric = ServiceCallMetric(
request_count_urn=monitoring_infos.API_REQUEST_COUNT_URN,
base_labels=labels)
# Get object state.
self._get_request = (
storage.StorageObjectsGetRequest(
bucket=self._bucket, object=self._name))
try:
metadata = self._get_object_metadata(self._get_request)
service_call_metric.call('ok')
except HttpError as http_error:
service_call_metric.call(http_error)
if http_error.status_code == 404:
raise IOError(errno.ENOENT, 'Not found: %s' % self._path)
else:
_LOGGER.error(
'HTTP error while requesting file %s: %s', self._path, http_error)
raise
self._size = metadata.size
# Ensure read is from file of the correct generation.
self._get_request.generation = metadata.generation
# Initialize read buffer state.
self._download_stream = io.BytesIO()
self._downloader = transfer.Download(
self._download_stream,
auto_transfer=False,
chunksize=self._buffer_size,
num_retries=20)
try:
self._client.objects.Get(self._get_request, download=self._downloader)
service_call_metric.call('ok')
except HttpError as e:
service_call_metric.call(e)
def _fetch_next_if_buffer_exhausted(self):
if not self.buffer or (self.buffer_start_position + len(self.buffer) <=
self.position):
bytes_to_request = min(self._remaining(), self.buffer_size)
self.buffer_start_position = self.position
retry_count = 0
while retry_count <= 10:
queue = Queue.Queue()
t = threading.Thread(target=self._fetch_to_queue,
args=(queue, self._get_segment,
(self.position, bytes_to_request)))
t.daemon = True
t.start()
try:
result, exn, tb = queue.get(timeout=self.segment_timeout)
except Queue.Empty:
logging.warning((
'Timed out fetching %d bytes from position %d of %s after %f '
'seconds; retrying...'), bytes_to_request,
self.position, self.path,
self.segment_timeout)
retry_count += 1
# Reinitialize download objects.
self.download_stream = cStringIO.StringIO()
self.downloader = transfer.Download(
self.download_stream,
auto_transfer=False,
chunksize=self.buffer_size)
self.client.objects.Get(self.get_request,
download=self.downloader)
continue
if exn:
logging.error((
'Exception while fetching %d bytes from position %d of %s: '
'%s\n%s'), bytes_to_request, self.position, self.path,
exn, tb)
raise exn
self.buffer = result
return
raise GcsIOError(
'Reached retry limit for _fetch_next_if_buffer_exhausted.')
