Function test_check_meta

dask/dataframe/tests/test_utils_dataframe.py:405–491

()

def test_check_meta():
    df = pd.DataFrame(
        {
            "a": ["x", "y", "z"],
            "b": [True, False, True],
            "c": [1, 2.5, 3.5],
            "d": [1, 2, 3],
            "e": pd.Categorical(["x", "y", "z"]),
            "f": pd.Series([1, 2, 3], dtype=np.uint64),
        }
    )
    meta = df.iloc[:0]

    # DataFrame metadata passthrough if correct
    assert check_meta(df, meta) is df
    # Series metadata passthrough if correct
    e = df.e
    assert check_meta(e, meta.e) is e
    # numeric_equal means floats and ints are equivalent
    d = df.d
    f = df.f
    assert check_meta(d, meta.d.astype("f8"), numeric_equal=True) is d
    assert check_meta(f, meta.f.astype("f8"), numeric_equal=True) is f
    assert check_meta(f, meta.f.astype("i8"), numeric_equal=True) is f

    # Series metadata error
    with pytest.raises(ValueError) as err:
        check_meta(d, meta.d.astype("f8"), numeric_equal=False)
    series = "pandas.core.series.Series" if not PANDAS_GE_300 else "pandas.Series"
    assert str(err.value) == (
        "Metadata mismatch found.\n"
        "\n"
        f"Partition type: `{series}`\n"
        "+----------+---------+\n"
        "|          | dtype   |\n"
        "+----------+---------+\n"
        "| Found    | int64   |\n"
        "| Expected | float64 |\n"
        "+----------+---------+"
    )

    # DataFrame metadata error
    meta2 = meta.astype({"a": "category", "d": "f8"})[["a", "b", "c", "d"]]
    df2 = df[["a", "b", "d", "e"]]
    with pytest.raises(ValueError) as err:
        check_meta(df2, meta2, funcname="from_delayed")
    frame = "pandas.core.frame.DataFrame" if not PANDAS_GE_300 else "pandas.DataFrame"

    if PANDAS_GE_300:
        string_type = "str   "  # space for alignment
    else:
        string_type = "object"

    exp = textwrap.dedent(
        f"""\
        Metadata mismatch found in `from_delayed`.

        Partition type: `{frame}`