utf8 to utf16 index map

utf8_to_utf16_map.ts

// ty @retr0id

const getUtf8Size = (str: string, index: number): number => {
	const code = str.codePointAt(index)!;

	if (code <= 0x7f) {
		return 1;
	} else if (code <= 0x7ff) {
		return 2;
	} else if (code <= 0xffff) {
		return 3;
	} else {
		return 4;
	}
};

const getUtf8To16Mapping = (str: string): number[] => {
	const map: number[] = [];
	const len = str.length;

	let u16pos = 0;

	while (u16pos < len) {
		const code = str.codePointAt(u16pos)!;

		for (let i = getUtf8Size(str, u16pos); i--; ) {
			map.push(u16pos);
		}

		u16pos += code > 0xffff ? 2 : 1;
	}

	return map;
};