Is it possible for s3.to_parquet to use partition_cols as defined in Glue Catalog?

[mattfysh]

I find that I need to pass in explicit partition_cols when calling s3.to_parquet, even though these have already been defined in Glue Catalog. Is it possible to make it so that awswrangler grabs the partition columns from Glue (which it already seems to be doing for data types)?

[kukushking]

The difficulty here is to distinguish between the case where you would like to create/update the Glue table definition with the new partitioning scheme because atm wr.s3.to_parquet keeps table definition in sync with the data, not vice versa.

In your case I would suggest to use wr.catalog.get_table() to retrieve partition keys from the Glue table instead

[mattfysh]

Ok cheers, this makes sense, although it would be nice to have an option to avoid multiple GetTable calls