stdlib, embrace codePoints! ```inline fun CharSequ...

# stdlib

[miha-x64]

stdlib, embrace codePoints!

inline fun CharSequence.forEachCodePoint(
    start: Int = 0, end: Int = length,
    action: (index: Int, codePoint: Int) -> Unit
) {
    var i = start
    var codePoint: Int
    while (i < end) {
        codePoint = Character.codePointAt(this, i);
        action(i, codePoint)
        i += Character.charCount(codePoint)
    }
}

[jw]

https://youtrack.jetbrains.com/issue/KT-30509

[Mark]

To avoid the ambiguity of whether the start/end/index are char offsets or codepoint indexes, I just use a sequence and then drop/take as necessary. Not as efficient, but fine in most cases.

fun CharSequence?.asCodePointSequence(): Sequence<Int> {
    return if (this.isNullOrEmpty()) {
        emptySequence()
    } else {
        var nextOffset = 0
        sequence {
            while (nextOffset < length) {
                val result = Character.codePointAt(this@asCodePointSequence, nextOffset)
                nextOffset += Character.charCount(result)
                yield(result)
            }
        }
    }
}

[miha-x64]

Java has similar .codePoints(): IntStream but it's not really useful. I'd better use IntIterator.

[Mark]

What are the pros/cons of building an IntIterator vs Sequence<Int>?

[miha-x64]

Iterator has direct next() method. This makes easy, for example, to wrap it and make iterator of words: you just call next() until space or punctuation encountered.
Making the same thing to Sequence/Stream is basically the same but requires more indirection because Sequence/Stream it is not an Iterator itself but a (spl)iterator factory.
Example: https://youtrack.jetbrains.com/issue/KT-38384