How to split files based on size(less than 20mb per file) when writing into S3

Hello,

I need to write files into S3 by keeping file Size(<20mb) ,rather than giving number of records batch limit using pipeline execute. My input is dynamic sometimes 1k records write file >20mb , sometimes 10k records is <20mb. We also have batches> 100k records as inputs. So how can we write files based on Size.

thanks in advance.