Missing 'boto3_session' in aws-data-wrangler/awswrangler/s3.py

[hjuhel-cdpq]

Describe the bug
Using s3.to_parquetwith mode overwrite_partitions fails since the call to delete_objects does not propagate the boto3_session

To Reproduce

• Try to create a dataset while providing a boto3 session

import awswrangler as wr
import pandas as pd
import boto3

# The custom boto3 session 
sess = boto3.Session(aws_access_key_id='<foo>', aws_secret_access_key='<bar>')

# The dataframe to write
df = pd.DataFrame({'one': [0,], 'two': ["spam"], 'three': [True]})

wr.s3.to_parquet(
    df=df,
    path='s3://path',
    dataset=True,
    mode="overwrite_partitions",
    partition_cols=["two"],
    boto3_session=sess
)

The previous snipet results in an authentication error, probably because of the missing boto3_session parameter in the delete call:

if mode == "overwrite_partitions":
    delete_objects(path=prefix, use_threads=use_threads

[igorborgest]

Hi @hjuhel-cdpq! Thank you very much for report it.

Will be fixed and released later today in the version 1.0.2.

[igorborgest]

Thanks @hjuhel-cdpq, this issue is resolved on version 1.0.2. Please, give feedback when possible.