AttributeError: 'Index' object has no attribute 'head' - Regression in 2.15.*

[mwoods-familiaris]

Describe the bug

Regression of #1188 in awswrangler 2.15.*

wr.s3.to_parquet fails when writing a DataFrame with an an ExtensionDType index resulting in error

AttributeError: 'Index' object has no attribute 'head'

Specifically, the failure happens when using pandas==1.4.* and awswrangler==2.15.*. This behavior is consistent on both pyarrow==5.0.0 and pyarrow==7.0.0. Appears to be a regression in the 2.15.0 release, as works fine when awswrangler==2.14.*

How to Reproduce

import pandas as pd
import awswrangler as wr


df = pd.DataFrame(
    {"col1": [1, 2, 3], "col2": [1, 2, 3]}, dtype=pd.Int64Dtype()
).set_index("col1")

wr.s3.to_parquet(
    df,
    path="s3://bucket/tmp.parquet",
    index=True,
    dataset=False,
)
# throws error

Expected behavior

Successful write of parquet file to S3 location

Your project

No response

Screenshots

No response

OS

Linux / Debian 10.12

Python version

3.9

AWS DataWrangler version

2.15.0, 2.15.1

Additional context

Pandas 1.4.2

[kukushking]

There's all sorts of strange issues when using extension types in indexes in pandas 1.4.* i.e. this as well.

Let's see if above fix helps with this particular case & doesn't break anything else

[updiversity]

Hi,

it looks like it continues to fail in the case the index has no name.

import pandas as pd
import awswrangler as wr

df = pd.DataFrame(
    {"col1": [1, 2, 3], "col2": [1, 2, 3]}, dtype=pd.Int64Dtype()
).set_index("col1")

df.index.name = None

wr.s3.to_parquet(
    df,
    path="s3://bucket/tmp.parquet",
    index=True,
    dataset=False,
)

# throws error
```

This is a common case when dealing with df generated with user-functions.