s3_organize.py

from datetime import datetime
import boto3


def list_bucket_contents(bucket_name):
    """return a list of contents in a bucket"""
    client = boto3.client('s3')
    paginator = client.get_paginator('list_objects')
    try:
        page_iterator = paginator.paginate(Bucket=bucket_name)
    except:
        print(bucket_name)
        return []

    def page2csv(page):
        contents = page['Contents']
        return [','.join([c['LastModified'].strftime('%Y-%m-%d %H:%M'),
                          c['Key'],
                          str(c['Size'])]) for c in contents]

    return sum([page2csv(p) for p in page_iterator], [])


if __name__ == '__main__':
    client = boto3.client('s3')
    response = client.list_buckets()
    bucket_names = [b['Name'] for b in response['Buckets']]
    bucket_names = ['foo']
    for bucket_name in bucket_names:
        with open(bucket_name + '_contents.csv', 'w') as f:
            contents = list_bucket_contents(bucket_name)
            f.write('\n'.join(contents))