"""Docstring."""

# AWS CLI sync is very fast when copying single dir
# this is a PoC that applies to many files not in the same subdir

import asyncio
import hashlib
import sys
import time

import aioboto3


async def main():
    """Entrypoint."""
    start = time.perf_counter()
    bucket = 'dev-pdumoulin'
    prefix = sys.argv[1]
    session = aioboto3.Session()
    async with session.client('s3') as client:
        keys = await list_bucket(client, bucket, prefix)
        results = await asyncio.gather(
            *[
                download(client, bucket, key)
                for key in keys
            ]
        )
    for result in results:
        print(result)
    print(f'Total time: {time.perf_counter() - start}')


async def list_bucket(client, bucket, prefix):
    """S3 list all objects."""
    keys = set()
    paginator = client.get_paginator('list_objects')
    async for page in paginator.paginate(Bucket=bucket, Prefix=prefix):
        for content in page['Contents']:
            key = content['Key']
            if not key.endswith('/'):
                keys.add(key)
    return keys


async def download(client, bucket, path):
    """S3 download operation."""
    chunk_size = 1024 * 64
    start = time.perf_counter()
    print('Downloading: ', bucket, path)
    md5_hash = hashlib.md5()
    file_data = await client.get_object(Bucket=bucket, Key=path)
    async for chunk in file_data['Body'].iter_chunks(chunk_size=chunk_size):
        md5_hash.update(chunk)
    print('Finished: ', bucket, path, f'{time.perf_counter() - start}')
    return (path, md5_hash.hexdigest(), file_data['ContentLength'])


if __name__ == '__main__':
    asyncio.run(main())