Question about Cloudtrail ingest processor for file.path

We've noticed while using filebeat to process logs from Cloudtrail that for some reason the processor looks at the previous S3 object and are unsure of the reasoning behind it:

  - rename:
      field: "json.previousDigestS3Object"
      target_field: "file.path"
      ignore_failure: true

GitHub

This is a bit counter-intuitive, as we initially thought that a field called file.path would contain the name of the current file instead of the previous file? Or is there something we are missing?