Find most recently uploaded file in a Backblaze B2 bucket (with optional prefix)

last_upload.py

import argparse
from datetime import datetime, MINYEAR, timezone

import boto3
from dotenv import load_dotenv


def list_objects(client, bucket_name, prefix=''):
    """
    Python generator to allow easy iteration over object versions, making
    API calls as necessary rather than reading the entire list into memory.

    Adapted from https://stackoverflow.com/a/54014862/33905

    :param client: a boto3 S3 client object
    :param bucket_name: the bucket name
    :param prefix: an optional prefix; defaults to ''
    :return: a single object version
    """
    s3_paginator = client.get_paginator('list_objects_v2')
    for page in s3_paginator.paginate(Bucket=bucket_name, Prefix=prefix):
        for version in page.get('Contents', ()):
            yield version


def main():
    parser = argparse.ArgumentParser(description='Find most recent upload to a bucket')

    parser.add_argument('bucket', type=str, nargs=1, help='a bucket name')
    parser.add_argument('prefix', type=str, nargs='?', default='', help='a prefix within the bucket')
    args = parser.parse_args()

    bucket = args.bucket[0]
    prefix = args.prefix[1:] if args.prefix.startswith('/') else args.prefix

    # Never put credentials in source code!
    # Create a file in this directory, named .env, with contents:
    # AWS_ACCESS_KEY_ID=your application key id
    # AWS_SECRET_ACCESS_KEY=your application key
    # AWS_REGION=region from your bucket endpoint, e.g. us-west-004
    # AWS_ENDPOINT_URL=your bucket endpoint, e.g. https://s3.us-west-004.backblazeb2.com
    load_dotenv()

    client = boto3.client('s3')

    max_last_modified_time = datetime(MINYEAR, 1, 1, tzinfo=timezone.utc)
    max_last_modified_key = ''
    count = 0
    for obj in list_objects(client, bucket, prefix=prefix):
        count += 1
        if obj['LastModified'] > max_last_modified_time:
            max_last_modified_time = obj['LastModified']
            max_last_modified_key = obj['Key']

    path = bucket if len(prefix) == 0 else f'{bucket}/{prefix}'
    if max_last_modified_key:
        print(f'Last upload to {path} was {max_last_modified_key} at {max_last_modified_time}')
    else:
        print(f'No files in {path}')

if __name__ == "__main__":
    main()