How do I write bytes to a binary column?

[mattfysh]

When attempting to pass a bytes object into a binary column, the bytes are converted into a string and appear in the database as (e.g.) "b'abc'"

Do I need to convert the bytes to a new type of object first, before calling s3.to_parquet?

[mattfysh]

This bug occurs only when using dtypes, either directly by passing in the parameter, or when retrieved from Glue

on

aws-sdk-pandas/awswrangler/_data_types.py

Lines 702 to 703 in cdb185f

	elif desired_type == "bytes":
	df[col] = df[col].astype("string").str.encode(encoding="utf-8").replace(to_replace={pd.NA: None})

A bytes type in pandas is object and when the desired type is bytes - for some reason - the column is being converted to a string and re-encoded into utf-8 ... causing a bytes array of abc (or 61 62 63) to be stringified and then encoded as b'abc' (or 62 27 61 62 63 27)