[ARROW-9787] [Python] pa.array.to_pandas(types_mapper=...) doesn't change the dtype

• Type: Bug
• Status: Closed
• Priority: Major
• Resolution: Duplicate
• Affects Version/s: 1.0.0
• Component/s: Python
• External issue URL: https://github.com/apache/arrow/issues/25834
• Assignee: Unassigned
• Reporter: quentin lhoest
• Created: 18/Aug/20 15:55
• Updated: 11/Jan/23 08:09
• Resolved: 18/Aug/20 16:20

Description

Hello !

It looks like it is possible to map arrow types to pandas types using the `types_mapper` argument in `to_pandas`. However it seems to work only for `pa.Table.to_pandas` and not `pa.array.to_pandas`.

Example to reproduce:

import pyarrow as pa
import pandas as pd

def types_mapper(dtype):
    print("Looking at dtype")
    if dtype == pa.int64():
        print("Changing dtype")
        return pd.Int32Dtype()

arr = pa.array([1,2,3], pa.int64())

print(pa.Table.from_arrays([arr], names=["col"]).to_pandas(types_mapper=types_mapper).col.dtype)
# Prints:
# Looking at dtype
# Changing dtype
# Int32

print(arr.to_pandas(types_mapper=types_mapper).dtype)
# Prints:
# int64

Issue Links

• duplicates: ARROW-9664 [Python] Array/ChunkedArray.to_pandas do not support types_mapper keyword