import os
from datetime import datetime, timedelta, timezone
import boto3


def main():
    aws_profile = os.getenv("AWS_PROFILE")
    if not aws_profile:
        raise ValueError("Missing variable AWS_PROFILE")
    bucket_name = os.getenv("S3_BUCKET")
    if not bucket_name:
        raise ValueError("Missing variable S3_BUCKET")
    days = os.getenv("DAYS")
    if not days:
        raise ValueError("Missing variable DAYS")
    days = int(days)
    assert days > 0

    date_threshold = datetime.now(timezone.utc) - timedelta(days=days)

    session = boto3.session.Session(profile_name=aws_profile)
    s3_resource = session.resource("s3")
    if not s3_resource.Bucket(bucket_name).creation_date:
        raise ValueError(f"Unknown bucket {bucket_name}")

    keys = []
    s3_client = session.client("s3")
    s3_paginator = s3_client.get_paginator("list_objects_v2")
    s3_pages = s3_paginator.paginate(Bucket=bucket_name)

    for s3_page in s3_pages:
        if "Contents" in s3_page:
            print(f"Found {len(s3_page['Contents'])} objects")
            for key_object in s3_page["Contents"]:
                if key_object["LastModified"] < date_threshold:
                    keys.append({
                        "Key": key_object["Key"],
                    })
                    if len(keys) > 999:
                        print(f"Will delete {len(keys)} objects")
                        s3_client.delete_objects(
                            Bucket=bucket_name,
                            Delete={
                                "Objects": keys,
                            },
                        )
                        keys = []

    if len(keys) > 1:
        print(f"Will delete {len(keys)} keys")
        s3_client.delete_objects(
            Bucket=bucket_name,
            Delete={
                "Objects": keys,
            },
        )


if __name__ == "__main__":
    main()