unicode

string.js

const forEachChar = (str, fn) => {
    if (str.length === 0) {
        return;
    }

    const iter = str[Symbol.iterator]();
    let it = iter.next();
    let prev = it.value;

    do {
        it = iter.next();

        const char = it.value;

        if (isCombiningMark(char)) {
            prev += char;
        } else {
            if (fn(prev) === false) {
                return;
            }

            prev = char;
        }
    } while (!it.done);
};
const charArrayLength = (str) => {
    let len = 0;

    forEachChar(str, () => {
        len++;
    });
    return len;
};
const toCharArray = (str) => {
    const arr = [];

    forEachChar(str, (char) => {
        arr.push(char);
    });
    return arr;
};
const transCharArrayOffset = (str, originOffset) => {
    let i = 0;

    forEachChar(str, (char) => {
        if (originOffset <= 0) {
            return false;
        }

        originOffset -= char.length;
        i++;
    });

    return i;
};
const REG_COMBINING_MARK = /[\u0300-\u036f\u1ab0-\u1aff\u1dc0-\u1dff\u20d0-\u20ff\ufe20-\ufe2f]/;
const isCombiningMark = (character) => {
    return REG_COMBINING_MARK.test(character);
};

toCharArray("1🧡2");
toCharArray("H̹̙̦̮͉̩̗̗ͧ̇̏̊̾Eͨ͆͒̆ͮ̃͏̷̮̣̫̤̣ ̵̞̹̻̀̉̓ͬ͑͡ͅCͯ̂͐͏̨̛͔̦̟͈̻O̜͎͍͙͚̬̝̣̽ͮ͐͗̀ͤ̍̀͢M̴̡̲̭͍͇̼̟̯̦̉̒͠Ḛ̛̙̞̪̗ͥͤͩ̾͑̔͐ͅṮ̴̷̷̗̼͍̿̿̓̽͐H̙̙̔̄͜");