cannot treat NumPy array of type

Error message

cannot treat NumPy array of type {arr.dtype} as indices

What it means

`_convert_np_ndarray_to_indices` accepts only arrays whose dtype kind is integer ('i' or 'u'). Boolean arrays are routed to a dedicated boolean-mask error; any other dtype (float, string, object) raises this TypeError because non-integer arrays cannot be positional indices.

Solutions

Cast to an integer dtype first: `s[arr.astype(np.uint32)]`.
Use integer division `//` instead of `/` when computing positions.
For label lookups use `.filter()` or a join rather than positional indexing.

Example fix

// before
arr = np.array([0.0, 2.0, 4.0])
sub = s[arr]  # TypeError: cannot treat NumPy array of type float64...
// after
sub = s[arr.astype(np.int64)]

Defensive patterns

Strategy: type-guard

Validate before calling

if isinstance(idx, np.ndarray) and idx.dtype.kind not in ('i', 'u'):
    if idx.dtype.kind == 'b':
        raise ValueError("use .filter() for boolean masks")
    idx = idx.astype(np.int64)

Type guard

def is_int_ndarray(x) -> bool:
    return isinstance(x, np.ndarray) and x.dtype.kind in ('i', 'u')

Try / catch

try:
    sub = s[np_idx]
except TypeError as e:
    if "as indices" in str(e):
        sub = s[np.asarray(np_idx).astype(np.int64)]
    else:
        raise

Prevention

Cast float index arrays to int before indexing.
Use `//` instead of `/` for position arithmetic.
Send boolean arrays to `.filter()`.

When it happens

Trigger: `s[np.array([0.0, 1.0])]`, `s[np.array(['a','b'])]`, or an index array produced by float division instead of integer division.

Common situations: Getting float indices from `np.linspace`, `/` division, or `np.where` on floats; loading indices from JSON which yields floats; passing string columns as selectors by mistake.

Related errors

AI-assisted analysis of pola-rs/polars@fe841f959e (2026-09-18). Data as JSON: /api/errors/79592b29a443e920. Report an issue: GitHub.

Appendix: source

Thrown at py-polars/src/polars/_utils/getitem.py:425

    #     to absolute indexes.
    if arr.ndim == 0:
        arr = np.atleast_1d(arr)
    if arr.ndim != 1:
        msg = "only 1D NumPy arrays can be treated as indices"
        raise TypeError(msg)

    idx_type = get_index_type()

    if len(arr) == 0:
        return pl.Series("", [], dtype=idx_type)

    # Numpy array with signed or unsigned integers.
    if arr.dtype.kind not in ("i", "u"):
        if arr.dtype.kind == "b":
            _raise_on_boolean_mask()
        else:
            msg = f"cannot treat NumPy array of type {arr.dtype} as indices"
            raise TypeError(msg)

    if idx_type == UInt32:
        if arr.dtype in {np.int64, np.uint64} and arr.max() >= U32_MAX:
            msg = "index positions should be smaller than 2^32"
            raise ValueError(msg)
        if arr.dtype == np.int64 and arr.min() < -U32_MAX:
            msg = "index positions should be greater than or equal to -2^32"
            raise ValueError(msg)

    if arr.dtype.kind == "i" and arr.min() < 0:
        if idx_type == UInt32:
            if arr.dtype in (np.int8, np.int16):
                arr = arr.astype(np.int32)
        else:
            if arr.dtype in (np.int8, np.int16, np.int32):
                arr = arr.astype(np.int64)

        # Update negative indexes to absolute indexes.

View on GitHub (pinned to fe841f959e)