get metadata downloads the entire blob

[yan-hic]

Not sure if it is by design but I have noticed that in order to read the metadata of a blob, the library actually downloads it, before parsing and rendering the metadata dict.
If this is the case, this is highly inefficient e.g. one needs to download a 2GB file to read the content_type or any custom metadata.

If confirmed, this is a showstopper for us.

Also, list_blobs only returns name so can't use it to read the metadata - which would be more efficient.

[guseggert]

download_metadata() just downloads the metadata, it calls https://cloud.google.com/storage/docs/json_api/v1/objects/get without alt=media.

resp_bytes = await client._download(bucket, key)
obj_size = json.loads(resp_bytes.decode())['size']
print("obj_size", obj_size)
print("bytes", len(resp_bytes))

results in

obj_size 1634987927
bytes 874