Function test_combine_first

dask/dataframe/tests/test_dataframe.py:1850–1876

()

def test_combine_first():
    df1 = pd.DataFrame(
        {
            "A": np.random.choice([1.0, 2.0, np.nan], 100),
            "B": np.random.choice(["a", "b", "nan"], 100),
        }
    )

    df2 = pd.DataFrame(
        {
            "A": np.random.choice([1.0, 2.0, 3.0], 100),
            "B": np.random.choice(["a", "b", "c"], 100),
        }
    )
    ddf1 = dd.from_pandas(df1, 4)
    ddf2 = dd.from_pandas(df2, 5)

    # DataFrame
    assert_eq(ddf1.combine_first(ddf2), df1.combine_first(df2))
    assert_eq(ddf1.combine_first(df2), df1.combine_first(df2))

    # Series
    assert_eq(ddf1.A.combine_first(ddf2.A), df1.A.combine_first(df2.A))
    assert_eq(ddf1.A.combine_first(df2.A), df1.A.combine_first(df2.A))

    assert_eq(ddf1.B.combine_first(ddf2.B), df1.B.combine_first(df2.B))
    assert_eq(ddf1.B.combine_first(df2.B), df1.B.combine_first(df2.B))


# This will likely break when the deprecation is enacted in Pandas 3.0