From 68f5db6d14789fe050ce6ee21adee6c1fb53cd70 Mon Sep 17 00:00:00 2001 From: e-belfer Date: Fri, 4 Oct 2024 13:07:00 -0400 Subject: [PATCH] Remove excess method in save_zenodo_metrics, update docstrings for make_request() method --- src/usage_metrics/helpers.py | 4 ++-- .../scripts/save_zenodo_metrics.py | 17 ++++++++--------- 2 files changed, 10 insertions(+), 11 deletions(-) diff --git a/src/usage_metrics/helpers.py b/src/usage_metrics/helpers.py index 31fe240..9e24e4e 100644 --- a/src/usage_metrics/helpers.py +++ b/src/usage_metrics/helpers.py @@ -161,10 +161,10 @@ def get_table_name_from_context(context: OutputContext) -> str: def make_request( url: str, headers: str | None = None, params: str | None = None, timeout: int = 100 ) -> requests.models.Response: - """Makes a request to the github api. + """Makes a request with some error handling. Args: - query (str): A github api request url. + query (str): A request url. headers (str): Header to include in the request. params (str): Params of request. timeout (int): Timeout of request (in seconds). diff --git a/src/usage_metrics/scripts/save_zenodo_metrics.py b/src/usage_metrics/scripts/save_zenodo_metrics.py index 0d27caa..ce9eee9 100644 --- a/src/usage_metrics/scripts/save_zenodo_metrics.py +++ b/src/usage_metrics/scripts/save_zenodo_metrics.py @@ -43,8 +43,12 @@ def check_empty_string(cls, doi: str): # noqa: N805 return -def get_zenodo_logs() -> pd.DataFrame(): - """Download JSONs of metrics for all versions in the Catalyst Cooperative Zenodo community.""" +def save_zenodo_logs() -> pd.DataFrame(): + """Get JSONs of Zenodo metrics for all Catalyst records and upload to GCS. + + Get metrics for all versions in the Catalyst Cooperative Zenodo community locally, + and then upload to the sources.catalyst.coop GCS bucket. + """ bucket_name = "pudl-usage-metrics-archives.catalyst.coop" client = storage.Client() bucket = client.get_bucket(bucket_name) @@ -69,17 +73,12 @@ def get_zenodo_logs() -> pd.DataFrame(): def upload_to_bucket( bucket: storage.Client.bucket, blob_name: str, data: pd.DataFrame ) -> None: - """Upload a gcp object.""" + """Upload a GCP object to a selected bucket.""" blob = bucket.blob(blob_name) blob.upload_from_string(data) logger.info(f"Uploaded {blob_name} to GCS bucket.") -def save_metrics(): - """Save Zenodo traffic metrics to google cloud bucket.""" - get_zenodo_logs() - - if __name__ == "__main__": - sys.exit(save_metrics()) + sys.exit(save_zenodo_logs())