[ARROW-15291] [C++][Python] Segfault in StructArray.to_numpy and to_pandas if it contains an ExtensionArray

• Type: Bug
• Status: Resolved
• Priority: Major
• Resolution: Fixed
• Affects Version/s: 6.0.1
• Fix Version/s: 8.0.0
• Component/s: C++, Python
• Labels: pull-request-available
• Environment: pyarrow 6.0.1, macbook pro
• External issue URL: https://github.com/apache/arrow/issues/30782
• Assignee: Alenka Frim
• Reporter: quentin lhoest
• Created: 10/Jan/22 18:06
• Updated: 11/Jan/23 11:36
• Resolved: 01/Mar/22 13:08

Description

Hi !

If you create a StructArray with an ExtensionArray in it, then both to_numpy and to_pandas  segfault in python:

import pyarrow as pa

class CustomType(pa.PyExtensionType):
    def __init__(self):
        pa.PyExtensionType.__init__(self, pa.binary())
    def __reduce__(self):
        return CustomType, ()

arr = pa.ExtensionArray.from_storage(CustomType(), pa.array([b"foo"]))
pa.StructArray.from_arrays([arr], ["name"]).to_numpy(zero_copy_only=False)
 

Thanks in advance for the help !

Issue Links

• links to: GitHub Pull Request #12505