numpy2-deprecation (NPY201)

Fix is sometimes available.

What it does

Checks for uses of NumPy functions and constants that were removed from the main namespace in NumPy 2.0.

Why is this bad?

NumPy 2.0 includes an overhaul of NumPy's Python API, intended to remove redundant aliases and routines, and establish unambiguous mechanisms for accessing constants, dtypes, and functions.

As part of this overhaul, a variety of deprecated NumPy functions and constants were removed from the main namespace.

The majority of these functions and constants can be automatically replaced by other members of the NumPy API or by equivalents from the Python standard library. With the exception of renaming numpy.byte_bounds to numpy.lib.array_utils.byte_bounds, all such replacements are backwards compatible with earlier versions of NumPy.

This rule flags all uses of removed members, along with automatic fixes for any backwards-compatible replacements.

Examples

import numpy as np

arr1 = [np.Infinity, np.NaN, np.nan, np.PINF, np.inf]
arr2 = [np.float_(1.5), np.float64(5.1)]
np.round_(arr2)

Use instead:

import numpy as np

arr1 = [np.inf, np.nan, np.nan, np.inf, np.inf]
arr2 = [np.float64(1.5), np.float64(5.1)]
np.round(arr2)