0.10.0 has improper handling of pd.NA values when the dtype is, e.g., float32[pyarrow]

[adamroyjones]

(This is in the same region as code as the AI-generated report #428.)

Here's a(n approximately) minimal reproducing example:

# test_reprex.py
import pytest
import pandas as pd
from tabulate import tabulate


def test_reprex():
    df = pd.DataFrame({"id": ["a"], "x": [pd.NA]}).astype(
        {"id": "string", "x": "float32[pyarrow]"}
    )
    with pytest.raises(TypeError):
        tabulate(df, maxcolwidths=10)

A stack trace:

.venv/lib/python3.11/site-packages/tabulate/__init__.py:2308: in tabulate
    list_of_lists = _wrap_text_to_colwidths(
.venv/lib/python3.11/site-packages/tabulate/__init__.py:1704: in _wrap_text_to_colwidths
    if cell == "" or _isnumber(cell)
       ^^^^^^^^^^
_ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _

>   ???
E   TypeError: boolean value of NA is ambiguous

This is with Python 3.11.15 and the following packages.

Package         Version
--------------- -----------
iniconfig       2.3.0
numpy           2.4.4
packaging       26.2
pandas          3.0.3
pluggy          1.6.0
pyarrow         24.0.0
pygments        2.20.0
pytest          9.0.3
python-dateutil 2.9.0.post0
ruff            0.15.13
six             1.17.0
tabulate        0.10.0

Please let me know if you need or want more information.