[theRizwan]

#349 made parser.className({ value }) escape its value, which fixed #182. parser.id({ value }) has the same gap:

parser.id({ value: "hover:bg" }).toString();   // "#hover:bg"

const id = parser.id();
id.value = "hover:bg";
id.toString();                                  // "#hover\\:bg"

#hover:bg parses back as the id hover followed by a :bg pseudo-class, so a selector built through the API comes out with a different meaning.

The fix

The same change #349 made for ClassName. The ID constructor runs its value through the escaping setter unless raws.value was given. In splitWord(), the parser removes the raws the constructor adds when it had none, as it does for class names, so parsed ids keep their original spelling, including Less such as #foo@{bar}.

Verification

The tests mirror the ones #349 added for classes: five values that need escaping (a colon, a leading digit, a dot, a backslash, a non-ASCII character), explicit raws.value, a plain identifier getting no raws, and parsed ids keeping their raws state when cloned. The five escaping tests fail on main.

npm test is green: lint, type check, and coverage gates.