Thanks for the PR @AndPuQing .

Was there a special case where values were wrongly set to 0 that is fixed with your function? Could you please explain why this fixes issue #3895?
I would also appreciate a test, at best using the example from the issue.

Thanks for the PR @AndPuQing .

Was there a special case where values were wrongly set to 0 that is fixed with your function? Could you please explain why this fixes issue #3895? I would also appreciate a test, at best using the example from the issue.

This PR fixes a bug where plot_text color appears as black.

The issue arises due to a change in the string representation of NumPy scalars introduced in NumPy 2.x. Specifically, this change affects how arrays are converted to strings. For reference, see the NumPy 2.0 release notes

Example

In NumPy 2.0:

>>> import numpy as np
>>> x = np.ones(4)
>>> tuple(x)
(np.float64(1.0), np.float64(1.0), np.float64(1.0), np.float64(1.0))

In this case, the CSS color fails to take effect because the string representations are invalid, resulting in the color being displayed as black.

In NumPy 1.x:

>>> import numpy as np
>>> x = np.ones(4)
>>> tuple(x)
(1.0, 1.0, 1.0, 1.0)

Here, the string representations are valid, and the CSS color is applied correctly.

Fix

The following code modification resolves the issue by converting NumPy arrays to Python lists using .tolist(), ensuring proper string representations:

def scale_color_to_255(color: tuple) -> tuple[int, int, int, float]:
    color = np.array(color).tolist() if isinstance(color[0], (np.float64, np.float32)) else color
    return tuple(int(c * 255) if i < 3 else c for i, c in enumerate(color))

This ensures the string representation is valid, and the colors are displayed correctly.

And sorry, because I used black format, there are many changes in code style.