Normalizer not removing accents

I'm using Normalizer followed by a regex to remove accents, but I'm getting back the same string with the accents still there.

import java.text.*

const val INPUT = "áéíóůø"
fun main() {
	println(Normalizer.normalize(INPUT, Normalizer.Form.NFC).replace("[\\p{InCombiningDiacriticalMarks}]+", ""))
	println(Normalizer.normalize(INPUT, Normalizer.Form.NFD).replace("[\\p{InCombiningDiacriticalMarks}]+", ""))
	println(Normalizer.normalize(INPUT, Normalizer.Form.NFKC).replace("[\\p{InCombiningDiacriticalMarks}]+", ""))
	println(Normalizer.normalize(INPUT, Normalizer.Form.NFKD).replace("[\\p{InCombiningDiacriticalMarks}]+", ""))
}

Output:

> áéíóůø
>
> áéíóůø
>
> áéíóůø
>
> áéíóůø

Kotlin playground: https://pl.kotl.in/62l6rUEUm

I've read a dozen questions here that say the way to strip accent marks in Java/Kotlin is to use java.text.Normalizer plus some minor variation of the above regular expression (sometimes without square brackets, sometimes without the plus). Even Apache Common's stripAccent function uses Normalizer for its implementation (but apparently handles to special characters too).

What am I doing wrong?

You did not make "[\\p{InCombiningDiacriticalMarks}]+" a Regex.

println(
    Normalizer.normalize(INPUT, Normalizer.Form.NFD)
        .replace("[\\p{InCombiningDiacriticalMarks}]+".toRegex(), "")
)

This produces:

aeiouø

Notice that the stroke in ø is not a diacritic mark. It can be decomposed to neither

• "o" and U+0338 COMBINING LONG SOLIDUS OVERLAY, or;
• "o" and U+0337 COMBINING SHORT SOLIDUS OVERLAY

You can see that these three all look a bit different: o̸øo̷

Also notice that there are two more blocks in Unicode that contains combining diacritics, called "Combining Diacritical Marks Extended" and "Combining Diacritical Marks Supplement". Consider including those in your regex too.