unicorn/prefer-number-coercion

parseFloat and parseInt read as many characters as they can and ignore the rest, so parseFloat("12px") is 12 and parseInt("1e3", 10) is 1. Number() converts the whole string, so bad input gives NaN, which you can check for.

The rule reports parseFloat(value) and parseInt(value, 10), including when they're written as Number.parseFloat and Number.parseInt. parseInt with no radix or with another radix isn't reported.

--fix doesn't change the code, because the result can differ. --fix-suggestions rewrites it to Number(value), or to Math.trunc(Number(value)) for parseInt.

Configure#

Off by default. To turn it on:

[lint.rules]
"unicorn/prefer-number-coercion" = "error" # or "warn"

Examples#

const price = parseFloat(input.value);
const quantity = Number.parseInt(form.quantity, 10);

const price = Number(input.value);
const quantity = Math.trunc(Number(form.quantity));

const color = parseInt(hex, 16);
const page = parseInt(query.page);

When not to use it#

The results differ. Number("") is 0 where parseFloat("") is NaN, and Number("12px") is NaN. If you parse strings with a unit or other text after the number, keep parseFloat.