unexpected error on `/\p{Script=Han}/gu`

[mesqueeb]

unexpected error on `/\p{Script=Han}/gu`

[verhovsky]

The issue is that there's a difference between JavaScript and PCRE syntax in these \p values and Regexr just parses them like PCRE, even when you tell it it's JavaScript.

PCRE allows

\pL
\p{L}
\p{Han}

Whereas JavaScript allows

\p{L}
\p{Script=Han}
\p{Letter}

(note it doesn't allow \pL)

In PCRE, some text by its own is treated as a Script/Script_Extension/General Category (and if it's not a matching possible value in any of the 3 then that's an error) and in JavaScript some text by its own is just as the General Category (and if it's not a valid General Category, i.e. \p{Han} is a Script not a General Category, then that's an error) so it has the extra Script=Han syntax which PCRE doesn't have, so the parser needs to know if it's parsing JavaScript or PCRE.

The error happens in here

regexr/dev/src/ExpressionLexer.js

Lines 621 to 642 in 1e38271

	parseUnicode(token, sub) {
	// unicodescript: \p{Cherokee}
	// unicodecat: \p{Ll} \pL
	// not: \P{Ll} \p{^Lu}
	let match = sub.match(/p\{\^?([^}]*)}/i), val = match && match[1], not = sub[0] === "P";
	if (!match && (match = sub.match(/[pP]([LMZSNPC])/))) { val = match[1]; }
	else { not = not !== (sub[2] === "^"); }
	token.l += match ? match[0].length : 1;
	token.type = "unicodecat";
	if (this._profile.unicodeScripts[val]) {
	token.type = "unicodescript";
	} else if (!this._profile.unicodeCategories[val]) {
	val = null;
	}
	if (not) { token.type = "not"+token.type; }
	if ((!this._profile.config.unicodenegated && sub[2] === "^") || !val) {
	token.error = {id: "unicodebad"}
	}
	token.value = val;
	token.clss = "charclass"
	return token;
	};