def credentials(self) -> Optional[auth_credentials.Credentials]:
"""Default credentials."""
if self._credentials:
return self._credentials
logger = logging.getLogger("google.auth._default")
logging_warning_filter = utils.LoggingFilter(logging.WARNING)
logger.addFilter(logging_warning_filter)
credentials, _ = google.auth.default()
logger.removeFilter(logging_warning_filter)
return credentials
def _retrieve_gcs_source_columns(
project: str,
gcs_csv_file_path: str,
credentials: Optional[auth_credentials.Credentials] = None,
) -> List[str]:
"""Retrieve the columns from a comma-delimited CSV file stored on Google Cloud Storage
Example Usage:
column_names = _retrieve_gcs_source_columns(
"project_id",
"gs://example-bucket/path/to/csv_file"
)
# column_names = ["column_1", "column_2"]
Args:
project (str):
Required. Project to initiate the Google Cloud Storage client with.
gcs_csv_file_path (str):
Required. A full path to a CSV files stored on Google Cloud Storage.
Must include "gs://" prefix.
credentials (auth_credentials.Credentials):
Credentials to use to with GCS Client.
Returns:
List[str]
A list of columns names in the CSV file.
Raises:
RuntimeError: When the retrieved CSV file is invalid.
"""
gcs_bucket, gcs_blob = utils.extract_bucket_and_prefix_from_gcs_path(
gcs_csv_file_path)
client = storage.Client(project=project, credentials=credentials)
bucket = client.bucket(gcs_bucket)
blob = bucket.blob(gcs_blob)
# Incrementally download the CSV file until the header is retrieved
first_new_line_index = -1
start_index = 0
increment = 1000
line = ""
try:
logger = logging.getLogger("google.resumable_media._helpers")
logging_warning_filter = utils.LoggingFilter(logging.INFO)
logger.addFilter(logging_warning_filter)
while first_new_line_index == -1:
line += blob.download_as_bytes(start=start_index,
end=start_index +
increment).decode("utf-8")
first_new_line_index = line.find("\n")
start_index += increment
header_line = line[:first_new_line_index]
# Split to make it an iterable
header_line = header_line.split("\n")[:1]
csv_reader = csv.reader(header_line, delimiter=",")
except (ValueError, RuntimeError) as err:
raise RuntimeError(
"There was a problem extracting the headers from the CSV file at '{}': {}"
.format(gcs_csv_file_path, err))
finally:
logger.removeFilter(logging_warning_filter)
return next(csv_reader)
