[ARROW-14686] [Python] Current endian decision for numpy does not work for built-in data type.

• Type: Improvement
• Status: Resolved
• Priority: Major
• Resolution: Fixed
• Affects Version/s: 7.0.0
• Fix Version/s: 7.0.0
• Component/s: Python
• Labels: pull-request-available
• External issue URL: https://github.com/apache/arrow/issues/30226
• Assignee: Kazuaki Ishizaki
• Reporter: Kazuaki Ishizaki
• Created: 12/Nov/21 06:58
• Updated: 11/Jan/23 08:41
• Resolved: 17/Nov/21 11:34

Description

Although the current NumPy converter code tries to detect endian against a given object, it does not work for a built-in object. This is because a build-in object may have native | byte order.

This issue causes test failure as follows:

FAILED pyarrow/tests/test_array.py::test_array_from_numpy_unicode - UnicodeDecodeError: 'utf-32-le' codec can't decode bytes in position 0-3: code point not in range(0x110000)
FAILED pyarrow/tests/test_array.py::test_array_from_strided - UnicodeDecodeError: 'utf-32-le' codec can't decode bytes in position 0-3: code point not in range(0x110000)
FAILED pyarrow/tests/test_array.py::test_array_from_numpy_str_utf8 - UnicodeDecodeError: 'utf-32-le' codec can't decode bytes in position 0-3: code point not in range(0x110000) 

 

Issue Links

• links to: GitHub Pull Request #11686
• links to: GitHub Pull Request #11687