BUG: pd.merge returns different column order if dataframe contains 0 rows

[rohanjain101]

Pandas version checks

• 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 main branch of pandas.

Reproducible Example

>>> df_a = pd.DataFrame({"Col0": pd.Series([], dtype="int32[pyarrow]"), "Col1": pd.Series([], dtype="int32[pyarrow]"), "Col2": pd.Series([], dtype="int32[pyarrow]")})
>>> df_b = pd.DataFrame({"Col0": pd.Series([], dtype="int32[pyarrow]"), "Col1": pd.Series([], dtype="int32[pyarrow]"), "Col2": pd.Series([], dtype="int32[pyarrow]")})
>>> pd.merge(df_a, df_b,left_on=["Col0", "Col2"], right_on=["Col1", "Col2"])
Empty DataFrame
Columns: [Col0_x, Col1_x, Col0_y, Col1_y, Col2]
Index: []

>>> df_a = pd.DataFrame({"Col0": pd.Series([1], dtype="int32[pyarrow]"), "Col1": pd.Series([1], dtype="int32[pyarrow]"), "Col2": pd.Series([1], dtype="int32[pyarrow]")})
>>> df_b = pd.DataFrame({"Col0": pd.Series([1], dtype="int32[pyarrow]"), "Col1": pd.Series([1], dtype="int32[pyarrow]"), "Col2": pd.Series([1], dtype="int32[pyarrow]")})
>>> pd.merge(df_a, df_b,left_on=["Col0", "Col2"], right_on=["Col1", "Col2"])
   Col0_x  Col1_x  Col2  Col0_y  Col1_y
0       1       1     1       1       1
>>>

Issue Description

Above example performs 2 joins, both input schemas are the same, but the returned schema by pd.merge is different if there are 0 rows vs atleast 1 row, input schema for both dfs was [Col0, Col1, Col2]. When there are 0 rows, pd.merge returns schema as [Col0_x, Col1_x, Col0_y, Col1_y, Col2]

When there's atleast 1 row, schema is returned as [Col0_x, Col1_x, Col2, Col0_y, Col1_y]. The position of Col2 differs if df contain 0 rows vs atleast 1 row.

Expected Behavior

I would expect output schema to always be returned as [Col0_x, Col1_x, Col2, Col0_y, Col1_y].

Installed Versions

INSTALLED VERSIONS

commit : e86ed37
python : 3.11.2.final.0
python-bits : 64
OS : Windows
OS-release : 10
Version : 10.0.22621
machine : AMD64
processor : Intel64 Family 6 Model 85 Stepping 7, GenuineIntel
byteorder : little
LC_ALL : None
LANG : None
LOCALE : English_United States.1252

pandas : 2.1.1
numpy : 1.25.2
pytz : 2023.3
dateutil : 2.8.2
setuptools : 68.0.0
pip : 23.2.1
Cython : None
pytest : 7.0.1
hypothesis : None
sphinx : 4.2.0
blosc : None
feather : None
xlsxwriter : None
lxml.etree : None
html5lib : None
pymysql : None
psycopg2 : None
jinja2 : 3.1.2
IPython : 8.14.0
pandas_datareader : None
bs4 : 4.12.2
bottleneck : None
dataframe-api-compat: None
fastparquet : None
fsspec : None
gcsfs : None
matplotlib : None
numba : None
numexpr : None
odfpy : None
openpyxl : None
pandas_gbq : None
pyarrow : 13.0.0
pyreadstat : None
pyxlsb : None
s3fs : None
scipy : None
sqlalchemy : None
tables : None
tabulate : 0.8.9
xarray : None
xlrd : None
zstandard : None
tzdata : 2023.3
qtpy : None
pyqt5 : None

[mroeschke]

Thanks for the report. Can reproduce this on main.

Noting that this is independent of the Arrow dtype in your example

[lukemanley]

Thanks for the report. Can reproduce this on main.

@mroeschke - are you actually able to reproduce on main? I cannot reproduce and I think #55028 probably fixed this.

[rohanjain101]

Thanks @lukemanley, yes this looks fixed.