How can I set custom index in opensearch serverless pipeline?

Hello, I am building a opensearch pipeline that reads csv data from s3 bucket (using sqs event) and stores that in a opensearch serverless collection. I am using the following configuration to create this and it works as expected. My problem is I want to create dynamic index - primary based on the filename of the parsed file. I tried passing an additional field in the SQS message, but data-prepper rejects that field.

My architecture currently is S3 -> SQS -> Opensearch serverless. In Opensearch pipeline, source is S3(SQS) with CSV processor and opensearch collection as the sink.

I was able to use {key} which is the S3 key, but my key is formatted as dt=2023-10/filename_202310.csv. I just want 202310 as the index. Is there a way to dynamically generate this?

Configuration (using data-prepper 2)

version: "2"
log-pipeline:
  source:
    s3:
      codec:
        newline:
      compression: "none"
      aws:
        region: "my-region"
        sts_role_arn: "my-role"
      acknowledgments: true
      scan:
        buckets:
          - bucket:
              name: "my-bucket"
  processor:
    - csv:
        source: "message"
        delimiter: "\t"
        delete_header: false
  sink:
    - opensearch:
        hosts: [ "my-serverless-host" ]
        aws:
          sts_role_arn: "my-role"
          region: "my-region"
          serverless: true
          serverless_options:
             network_policy_name: "my-network-policy"
        index: "vector_index" <--- want to make this dynamic, not sure how. 

https://opensearch.org/docs/latest/data-prepper/pipelines/configuration/sinks/opensearch/