BUG: cannot write xml with nullable dtypes

[dmyersturnbull]

• I have checked that this issue has not already been reported.

• I have confirmed this bug exists on the latest version of pandas.

• I have confirmed this bug exists on the master branch of pandas.

Reproducible Example

import logging
import numpy as np
import pandas as pd
import pytest

logger = logging.getLogger("test")

class XmlTest:

    def test_to_xml_dtypes(self):
        dtypes = [
            pd.StringDtype(),
            pd.BooleanDtype(),
            pd.Float64Dtype(),
            pd.Float32Dtype(),
            pd.Int64Dtype(),
            pd.UInt64Dtype(),
            pd.Int32Dtype(),
            pd.UInt32Dtype(),
            pd.Int16Dtype(),
            pd.UInt16Dtype(),
            pd.Int8Dtype(),
            pd.UInt8Dtype(),
            pd.StringDtype(),
        ]
        data = [
            pd.Series({"abc": 1, "xyz": pd.NA}),
            pd.Series({"abc": pd.NA, "xyz": 0}),
        ]
        failed = {}
        for dtype in dtypes:
            df = pd.DataFrame(data).astype(dtype)
            try:
                df.to_xml()
            except TypeError as e:
                logger.error(dtype, exc_info=True)
                failed[str(dtype)] = str(e)
        assert failed == [], f"Failed on dtypes: {failed}"

if __name__ == "__main__":
    pytest.main()

Issue Description

to_xml fails with nullable dtypes provided by Pandas like Float64Dtype. Example from a test run in CI; it fails on Ubuntu too.

Expected Behavior

Output valid xml that can be read with read_xml to get approximately the same DataFrame.

Installed Versions

INSTALLED VERSIONS ------------------ commit : f00ed8f python : 3.9.5.final.0 python-bits : 64 OS : Windows OS-release : 10 Version : 10.0.19042 machine : AMD64 processor : Intel64 Family 6 Model 165 Stepping 2, GenuineIntel byteorder : little LC_ALL : None LANG : None LOCALE : English_United States.1252

pandas : 1.3.0
numpy : 1.21.2
pytz : 2021.1
dateutil : 2.8.1
pip : 21.1.1
setuptools : 52.0.0.post20210125
Cython : None
pytest : 6.2.4
hypothesis : None
sphinx : None
blosc : None
feather : None
xlsxwriter : None
lxml.etree : 4.6.3
html5lib : 1.1
pymysql : 1.0.2
psycopg2 : None
jinja2 : None
IPython : None
pandas_datareader: None
bs4 : 4.9.3
bottleneck : None
fsspec : 2021.07.0
fastparquet : None
gcsfs : None
matplotlib : 3.4.2
numexpr : None
odfpy : None
openpyxl : 3.0.7
pandas_gbq : None
pyarrow : 5.0.0
pyxlsb : 1.0.8
s3fs : None
scipy : 1.7.0
sqlalchemy : None
tables : None
tabulate : 0.8.9
xarray : None
xlrd : 2.0.1
xlwt : None
numba : None
None

[ParfaitG]

Conversion of nullable types in read_xml is not yet supported. Good item to add to IO XML enhancements tracker. The best avenue I see would be in the _data_to_frame method that uses TextParser from pandas.io.parsers to infer data types. Possibly, only this external parser (shared by other IO methods) would need to be adjusted to infer nullable types.

As workaround, since all text nodes in XML documents are actually text types you can convert columns to object types before calling to_xml().

Currently, your reproducible example outputs the following:

df
    abc   xyz
0     1  <NA>
1  <NA>     0

Converting all columns to object types then allows to_xml() to run without error:

xml_str = df.astype('str').to_xml(index=False)
print(xml_str)
<?xml version='1.0' encoding='utf-8'?>
<data>
  <row>
    <abc>1</abc>
    <xyz>&lt;NA&gt;</xyz>
  </row>
  <row>
    <abc>&lt;NA&gt;</abc>
    <xyz>0</xyz>
  </row>
</data>

However, reading back in requires users to convert to needed types. Floats are generated here since it allows NaN values:

new_df = pd.read_xml(xml_str)

new_df
   abc  xyz
0  1.0  NaN
1  NaN  0.0