pola-rs/polars · error · InvalidArgument

unsupported data type

Error message

unsupported data type: {dtype}

What it means

_instantiate_flat_dtype in strategies/dtype.py draws a concrete flat dtype from a DataTypeClass/selection; dtypes it cannot instantiate fall through to the final else and raise InvalidArgument. It is the flat-dtype instantiation path used by _flat_dtypes and the inner strategy machinery.

Solutions

1. Restrict generated dtypes to supported flat ones via the `extra_dtypes`/`blocked_dtypes`-style selection or by passing explicit dtypes.
2. Update strategies/dtype.py to handle the new dtype in _instantiate_flat_dtype.
3. Ensure nested dtypes go through _instantiate_nested_dtype / _nested_dtypes, not the flat path.

Example fix

// before
series(dtype=MyCustomDtype)
// after
series(dtype=pl.Int64)

FLAT = {pl.Int8, pl.Int16, pl.Int32, pl.Int64, pl.UInt8, pl.UInt16, pl.UInt32, pl.UInt64, pl.Float32, pl.Float64, pl.String, pl.Boolean, pl.Date, pl.Time}
if not any(dtype == d for d in FLAT):
    raise ValueError(f"{dtype} is not an instantiable flat dtype here")

def is_flat_supported(dtype) -> bool:
    return not dtype.is_nested() and dtype not in (pl.Unknown,)

from polars.exceptions import InvalidArgument
try:
    dt = instantiate(dtype)
except InvalidArgument as e:
    dt = pl.Int64  # fallback default for the test pool

When it happens

Trigger: Hypothesis drawing a dtype class that _instantiate_flat_dtype does not recognize (e.g. a nested-only or plugin dtype appearing in the flat-dtype pool); passing an unsupported dtype class into series()/instantiate helpers that route through the flat path.

Common situations: Custom or third-party dtypes used with parametric testing; polars version drift introducing a dtype not yet covered in dtype.py; accidentally requesting nested dtypes through a flat-dtype entry point.

Related errors

• map keys cannot be a nested dtype, got
• unsupported data type
• map keys cannot be Null; specify a key type, e.g…
• polars.testing.parametric requires the 'hypothesis'…
• activate dtype

AI-assisted analysis of pola-rs/polars@fe841f959e (2026-09-18). Data as JSON: /api/errors/568d47e06a3938da. Report an issue: GitHub.

Appendix: source

Thrown at py-polars/src/polars/testing/parametric/strategies/dtype.py:272

        return Datetime(time_unit, time_zone)
    elif dtype == Duration:
        time_unit = draw(_time_units())
        return Duration(time_unit)
    elif dtype == Categorical:
        return Categorical()
    elif dtype == Enum:
        n_categories = draw(
            st.integers(min_value=1, max_value=_DEFAULT_ENUM_CATEGORIES_LIMIT)
        )
        categories = [f"c{i}" for i in range(n_categories)]
        return Enum(categories)
    elif dtype == Decimal:
        precision = draw(st.integers(min_value=1, max_value=38) | st.none())
        scale = draw(st.integers(min_value=0, max_value=precision or 38))
        return Decimal(precision, scale)
    else:
        msg = f"unsupported data type: {dtype}"
        raise InvalidArgument(msg)


@st.composite
def _nested_dtypes(
    draw: DrawFn,
    inner: SearchStrategy[DataType],
    allowed_dtypes: Sequence[PolarsDataType] | None = None,
    excluded_dtypes: Sequence[PolarsDataType] | None = None,
    *,
    allow_time_zones: bool = True,
) -> DataType:
    """Create a strategy for generating nested Polars :class:`DataType` objects."""
    if allowed_dtypes is None:
        allowed_dtypes = _NESTED_DTYPES
    if excluded_dtypes is None:
        excluded_dtypes = []

    dtype = draw(st.sampled_from(allowed_dtypes))

View on GitHub (pinned to fe841f959e)