mcalc/modules/lexer.js

function tokenize(expr) {
	// NOTE: not handling whitespace
	// NOTE: currently ends early if string doesn't match token

	// TODO: it would probably be beneficial to create some sort of
	// function token instead of updating strings in several different
	// places
	const regexp = /[0-9]+|[-+*/^]|\(|\)|sqrt|ord/gy;
	const matches = expr.matchAll(regexp);
	
	const tokens = [];
	for (const match of matches) {
		if (/[0-9]+/.test(match[0])) {
			tokens.push(BigInt(match[0]));
		} else if (/[-+*^/]/.test(match[0])) {
			tokens.push(match[0])
		} else if (match[0] === "(") {
			tokens.push("(");
		} else if (match[0] === ")") {
			tokens.push(")");
		} else if (match[0] === "sqrt") {
			tokens.push("sqrt")
		} else if (match[0] === "ord") {
			tokens.push("ord")
		}
	}

	return tokens;
}

export { tokenize };