Number.toLocaleString in pt-BR fails in vitest

Question

I've defined a function toCurrency to convert string or number to locale strings:

function toCurrency(
    value: string | number,
    locale: string = "pt-BR",
    currency: string = "BRL",
    options?: Intl.NumberFormatOptions
) {
    return Number(value).toLocaleString(locale, {
        maximumFractionDigits: 2,
        style: "currency",
        currency: currency,
        ...options,
    });
}

and a test to validate it:

describe("toCurrency", () => {
    test("toCurrency defined", async () => {
        expect(toCurrency).toBeDefined();
    });

    test("toCurrency", () => {
        const result = toCurrency(1, "pt-BR", "BRL");
        expect(result).toEqual("R$ 1,00");
    });
    
    test("toCurrency USD", () => {
        const result = toCurrency(1, "en-US", "USD");
        expect(result).toEqual("$1.00");
    });
});

But the tests fails only for pt-BR:

Answer 1

The Node.js version you're running may have a different version of the CLDR data than your test expects, and so you get a different result for toLocaleString.

Alternately, if you haven't written your test specifically against CLDR data, your test expects e.g. a normal 0x20 space between R$ and the number, while chances are the character as formatted is actually e.g. a non-breaking space (\xA0):

> function toCurrency(
...     value,
...     locale = "pt-BR",
...     currency = "BRL"
... ) {
...     return Number(value).toLocaleString(locale, {
...         maximumFractionDigits: 2,
...         style: "currency",
...         currency,
...     });
... }
> toCurrency(1)
'R$ 1,00'
> [...toCurrency(1)].map(c => c.codePointAt(0))
[
  82, 36, 160, 49,
  44, 48,  48
]

Note how the third codepoint is 160 – that's 0xA0.