How to get cost reports for prefixes in AWS S3

Question

I have tagged unversioned data stored in AWS S3 in the following form:

<bucket>/foo/
             item-1
             item-2
             item-3 (inserted, read, then deleted)
        /bar/
             item-4
             item-5

There is one bucket, several prefixes like foo and bar, and many items using those prefixes. Items are inserted, read, updated, and deleted daily. I would like to get a cost breakdown in S3, using prefixes, for example

+--------------+---------+---------------+
|     Date     | Cost, $ | AWS S3 Prefix |
+--------------+---------+---------------+
| 25-Jan-2024  |      11 | <bucket>/foo/*|
| 25-Jan-2024  |      15 | <bucket>/bar/*|
+--------------+---------+---------------+

I've looked at using Cost Allocation tags but that seems to be limited to a bucket level. I cannot get the cost breakdown based on differently-tagged resources inside the bucket. The CUR reports also don't group S3 resources using prefixes. Both cost allocation tags and CUR allow me to get this data:

+--------------+---------+---------------+
|     Date     | Cost, $ | AWS S3 Prefix |
+--------------+---------+---------------+
| 25-Jan-2024  |      11 | <bucket>      |
+--------------+---------+---------------+

How do I programmatically get the costs of resources in AWS S3 grouped by a prefix?