python脚本分批次删除指定时间范围内指定路径下的s3的历史version

import boto3
from datetime import datetime, timedelta

import pytz as pytz


def split_list(lst, batch_size):
    """
    将列表分割成指定大小的批次
    """
    for i in range(0, len(lst), batch_size):
        yield lst[i:i + batch_size]

def list_versions_with_time_range(bucket_name, prefix, start_time, end_time):
    """
    列出指定存储桶和前缀下特定时间范围内的所有对象版本
    """
    s3 = boto3.client('s3',
                      aws_access_key_id='XXX',
                      aws_secret_access_key='XX',
                      endpoint_url='http://s3.XXX',
                      region_name='XXXXXXX',
                      use_ssl=False)
    versions = []

    response = s3.list_object_versions(Bucket=bucket_name, Prefix=prefix)

    for version in response.get('Versions', []):
        last_modified = version['LastModified']
        if start_time <= last_modified <= end_time:
            versions.append({'Key': version['Key'], 'VersionId': version['VersionId']})

    while response.get('NextKeyMarker') and response.get('NextVersionIdMarker'):
        response = s3.list_object_versions(
            Bucket=bucket_name,
            Prefix=prefix,
            KeyMarker=response['NextKeyMarker'],
            VersionIdMarker=response['NextVersionIdMarker']
        )

        for version in response.get('Versions', []):
            last_modified = version['LastModified']
            if start_time <= last_modified <= end_time:
                versions.append({'Key': version['Key'], 'VersionId': version['VersionId']})

    return versions

def delete_objects(bucket_name, objects):
    """
    分批次删除对象
    """
    s3 = boto3.client('s3',
                      aws_access_key_id='XXXXXX',
                      aws_secret_access_key='XXX',
                      endpoint_url='http://s3.',
                      region_name='XX',
                      use_ssl=False)

    # 将对象列表分割成较小的批次
    batches = split_list(objects, 1000)  # 将每个批次的大小设为 1000

    for batch in batches:
        # 构建删除对象请求
        delete_list = [{'Key': obj['Key'], 'VersionId': obj['VersionId']} for obj in batch]

        # 执行删除对象操作
        response = s3.delete_objects(
            Bucket=bucket_name,
            Delete={'Objects': delete_list}
        )

        # 处理响应
        print("Deleted objects:", response.get('Deleted', []))
        print("Errors:", response.get('Errors', []))

# 替换为你的存储桶名称和指定目录路径
bucket_name = 'XXX'
directory_path = 'XXXX'

utc = pytz.utc
start_time = utc.localize(datetime(2023, 1, 1))  # 开始时间
end_time = utc.localize(datetime(2024, 2, 12))  # 结束时间

# 列出特定时间范围内的对象版本
versions_to_delete = list_versions_with_time_range(bucket_name, directory_path, start_time, end_time)
print(versions_to_delete)
# 分批次删除对象版本
delete_objects(bucket_name, versions_to_delete)