Missing name in index of series created from DataArray.to_series() after reindex

[yhlam]

What happened:

If a DataArray is reindex() with a list, the Series created from DataArray.to_series() has an index with no name in the reindexed dimension. It was working for 2+ dimension array in 0.16.2 (i.e. the Series has MultiIndex instead of Index), but 0.17.0 doesn't work. For 1 dimension array, both 0.16.2 and 0.17.0 are not working as expected.

What you expected to happen:

I expect the Index.names of Series created from DataArray.to_series() contains the name of the dimensions of the DataArray, no matter reindex() was called on the DataArray before or not.

Minimal Complete Verifiable Example:

import numpy as np
import pytest
import xarray as xr


def test_multiindex():
    array = xr.DataArray(
        np.arange(12).reshape((4, 3)),
        coords=[
            ("foo", ["a", "b", "c", "d"]),
            ("bar", ["x", "y", "z"]),
        ]
    )
    assert array.to_series().index.names == ["foo", "bar"]


def test_multiindex_reindex():
    array = xr.DataArray(
        np.arange(12).reshape((4, 3)),
        coords=[
            ("foo", ["a", "b", "c", "d"]),
            ("bar", ["x", "y", "z"]),
        ]
    )
    series = array.reindex(bar=["x", "y"]).to_series()
    assert series.index.names == ["foo", "bar"]


def test_index_str():
    array = xr.DataArray(np.arange(4), coords=[("foo", ["a", "b", "c", "d"])])
    assert array.to_series().index.names == ["foo"]


def test_index_str_reindex():
    array = xr.DataArray(np.arange(4), coords=[("foo", ["a", "b", "c", "d"])])
    series = array.reindex(foo=["a", "b", "c"]).to_series()
    assert series.index.names == ["foo"]


def test_index_int():
    array = xr.DataArray(np.arange(4), coords=[("foo", [1, 2, 3, 4])])
    assert array.to_series().index.names == ["foo"]


def test_index_int_reindex():
    array = xr.DataArray(np.arange(4), coords=[("foo", [1, 2, 3, 4])])
    series = array.reindex(foo=[1, 2, 3]).to_series()
    assert series.index.names == ["foo"]

Version 0.16.2:

It works as expected for MultiIndex. However, the index name is missing after reindex for 1 dimension array.

================================================= test session starts =================================================
platform win32 -- Python 3.8.6, pytest-6.2.2, py-1.10.0, pluggy-0.13.1
plugins: cov-2.11.1
collected 6 items

tests\test_to_series_index.py ...F.F                                                                             [100%]

====================================================== FAILURES =======================================================
_______________________________________________ test_index_str_reindex ________________________________________________

    def test_index_str_reindex():
        array = xr.DataArray(np.arange(4), coords=[("foo", ["a", "b", "c", "d"])])
        series = array.reindex(foo=["a", "b", "c"]).to_series()
>       assert series.index.names == ["foo"]
E       AssertionError: assert FrozenList([None]) == ['foo']
E         At index 0 diff: None != 'foo'
E         Use -v to get the full diff

tests\test_to_series_index.py:39: AssertionError
_______________________________________________ test_index_int_reindex ________________________________________________

    def test_index_int_reindex():
        array = xr.DataArray(np.arange(4), coords=[("foo", [1, 2, 3, 4])])
        series = array.reindex(foo=[1, 2, 3]).to_series()
>       assert series.index.names == ["foo"]
E       AssertionError: assert FrozenList([None]) == ['foo']
E         At index 0 diff: None != 'foo'
E         Use -v to get the full diff

tests\test_to_series_index.py:50: AssertionError
=============================================== short test summary info ===============================================
FAILED tests/test_to_series_index.py::test_index_str_reindex - AssertionError: assert FrozenList([None]) == ['foo']
FAILED tests/test_to_series_index.py::test_index_int_reindex - AssertionError: assert FrozenList([None]) == ['foo']
============================================= 2 failed, 4 passed in 0.61s =============================================

Version 0.17.0:

Index name is missing after reindex on both single level Index and MultiIndex.

================================================= test session starts =================================================
platform win32 -- Python 3.8.6, pytest-6.2.2, py-1.10.0, pluggy-0.13.1
plugins: cov-2.11.1
collected 6 items

tests\test_to_series_index.py .F.F.F                                                                             [100%]

====================================================== FAILURES =======================================================
_______________________________________________ test_multiindex_reindex _______________________________________________

    def test_multiindex_reindex():
        array = xr.DataArray(
            np.arange(12).reshape((4, 3)),
            coords=[
                ("foo", ["a", "b", "c", "d"]),
                ("bar", ["x", "y", "z"]),
            ]
        )
        series = array.reindex(bar=["x", "y"]).to_series()
>       assert series.index.names == ["foo", "bar"]
E       AssertionError: assert FrozenList(['foo', None]) == ['foo', 'bar']
E         At index 1 diff: None != 'bar'
E         Use -v to get the full diff

tests\test_to_series_index.py:28: AssertionError
_______________________________________________ test_index_str_reindex ________________________________________________

    def test_index_str_reindex():
        array = xr.DataArray(np.arange(4), coords=[("foo", ["a", "b", "c", "d"])])
        series = array.reindex(foo=["a", "b", "c"]).to_series()
>       assert series.index.names == ["foo"]
E       AssertionError: assert FrozenList([None]) == ['foo']
E         At index 0 diff: None != 'foo'
E         Use -v to get the full diff

tests\test_to_series_index.py:39: AssertionError
_______________________________________________ test_index_int_reindex ________________________________________________

    def test_index_int_reindex():
        array = xr.DataArray(np.arange(4), coords=[("foo", [1, 2, 3, 4])])
        series = array.reindex(foo=[1, 2, 3]).to_series()
>       assert series.index.names == ["foo"]
E       AssertionError: assert FrozenList([None]) == ['foo']
E         At index 0 diff: None != 'foo'
E         Use -v to get the full diff

tests\test_to_series_index.py:50: AssertionError
=============================================== short test summary info ===============================================
FAILED tests/test_to_series_index.py::test_multiindex_reindex - AssertionError: assert FrozenList(['foo', None]) == [...
FAILED tests/test_to_series_index.py::test_index_str_reindex - AssertionError: assert FrozenList([None]) == ['foo']
FAILED tests/test_to_series_index.py::test_index_int_reindex - AssertionError: assert FrozenList([None]) == ['foo']
============================================= 3 failed, 3 passed in 0.61s =============================================

Environment:

Output of xr.show_versions() for 0.16.2

commit: None
python: 3.8.6 (tags/v3.8.6:db45529, Sep 23 2020, 15:52:53) [MSC v.1927 64 bit (AMD64)]
python-bits: 64
OS: Windows
OS-release: 10
machine: AMD64
processor: Intel64 Family 6 Model 158 Stepping 13, GenuineIntel
byteorder: little
LC_ALL: None
LANG: None
LOCALE: English_United States.1252
libhdf5: None
libnetcdf: None

xarray: 0.16.2
pandas: 1.2.3
numpy: 1.20.1
scipy: None
netCDF4: None
pydap: None
h5netcdf: None
h5py: None
Nio: None
zarr: None
cftime: None
nc_time_axis: None
PseudoNetCDF: None
rasterio: None
cfgrib: None
iris: None
bottleneck: None
dask: None
distributed: None
matplotlib: None
cartopy: None
seaborn: None
numbagg: None
pint: None
setuptools: 53.0.0
pip: 21.0.1
conda: None
pytest: 6.2.2
IPython: 7.21.0
sphinx: None

Output of xr.show_versions() for 0.17.0

commit: None
python: 3.8.6 (tags/v3.8.6:db45529, Sep 23 2020, 15:52:53) [MSC v.1927 64 bit (AMD64)]
python-bits: 64
OS: Windows
OS-release: 10
machine: AMD64
processor: Intel64 Family 6 Model 158 Stepping 13, GenuineIntel
byteorder: little
LC_ALL: None
LANG: None
LOCALE: English_United States.1252
libhdf5: None
libnetcdf: None

xarray: 0.17.0
pandas: 1.2.3
numpy: 1.20.1
scipy: None
netCDF4: None
pydap: None
h5netcdf: None
h5py: None
Nio: None
zarr: None
cftime: None
nc_time_axis: None
PseudoNetCDF: None
rasterio: None
cfgrib: None
iris: None
bottleneck: None
dask: None
distributed: None
matplotlib: None
cartopy: None
seaborn: None
numbagg: None
pint: None
setuptools: 53.0.0
pip: 21.0.1
conda: None
pytest: 6.2.2
IPython: 7.21.0
sphinx: None

[max-sixty]

Thanks for the excellent example @yhlam .

We'd definitely take a PR to fix this. We're also working on some larger index changes, though they are still some way off.

[sjvrijn]

The example tests by @yhlam currently pass in xarray: 2023.2.1.dev7+g21d86450. Using git bisect, it seems like this issue was fixed as part of the explicit indexes PR #5692. I guess that means this issue can be closed?

[dcherian]

Thanks @sjvrijn