jorenham

Rank-typing coverage of the public NumPy API.

numpy

Function	Progress	Effort	Limit	Uses (x1000)
array	●			20,218
arange	●			10,060
zeros[^shapearg]	●			9,716
ones[^shapearg]	●			6,676

jorenham

1 -> 1

ufunc	identity	arg	out
isnat		Mm	?
signbit		efdg	?
isfinite		?bBhHiIlLqQefdgFDGmM	?
isinf		?bBhHiIlLqQefdgFDGmM	?
isnan		?bBhHiIlLqQefdgFDGmMT	?
bitwise_count		bBhHiIlLqQO	BO

jorenham

jorenham

import subprocess
from pathlib import Path

# see https://github.com/astral-sh/ruff/pull/18224
from ruff.__main__ import (  # type: ignore[import-untyped]  # pyright: ignore[reportMissingTypeStubs]
    find_ruff_bin,  # noqa: PLC2701
)


def ruff_format(source: str) -> str:

jorenham

original post: scipy/scipy-stubs#153 (comment)

---

Annotating a function whose return type depends on an optional positional- or keyword-only parameter in Python

Overloads are roughly similar to pattern matching[^1], but from the perspective of the caller.

Consider this function for example

RandallPittmanOrSt

"""nptypes.py - Various useful types and type aliases"""

# FUTURE: Integrate optype

from typing import Any, Literal, Protocol, TypeAlias, TypeVar

import numpy as np
import numpy.typing as npt
from typing_extensions import TypeIs

jorenham

Without scipy-stubs	With scipy-stubs

• VSCode 1.96.2
• PyLance 2024.12.100
• pyright == 1.1.391
• scipy == 1.15.0

wolph

#!/usr/bin/env python3

import argparse
import configparser
import os
import re
import shlex
import sys

from typing import Any

twolfson

gist-python-templating-evaluation

wolph

#!/usr/bin/env python -u
import os
import re
import sys
import itertools
import contextlib
import subprocess

sys.stdout.reconfigure(line_buffering=True)