Support unicode identifiers

Author: Diogo Franco (Kovensky)

lib/parse.js

@@ -1,6 +1,7 @@
 "use strict";
 
 var Parser = require("fastparse");
+var regexpu = require("regexpu-core");
 
 function unescape(str) {
 	return str.replace(/\\(.)/g, "$1");
@@ -164,11 +165,19 @@ function bracketEnd(match) {
 	this.token.content += match;
 }
 
-var parser = new Parser({
-	selector: {
-		"/\\*([\\s\\S]*?)\\*/": commentMatch,
-		"\\.((?:\\\\.|[A-Za-z_\\-])(?:\\\\.|[A-Za-z_\\-0-9])*)": typeMatch("class"),
-		"#((?:\\\\.|[A-Za-z_\\-])(?:\\\\.|[A-Za-z_\\-0-9])*)": typeMatch("id"),
+function getSelectors() {
+	// The assignment here is split to preserve the property enumeration order.
+	var selectors = {
+		"/\\*([\\s\\S]*?)\\*/": commentMatch
+	};
+	// https://www.w3.org/TR/CSS21/syndata.html#characters
+	// 4.1.3: identifiers (...) can contain only the characters [a-zA-Z0-9] and
+	// ISO 10646 characters U+00A0 and higher, plus the hyphen (-) and the underscore (_)
+	//
+	// 10ffff is the maximum allowed in current Unicode
+	selectors[regexpu("\\.((?:\\\\.|[A-Za-z_\\-\\u{00a0}-\\u{10ffff}])(?:\\\\.|[A-Za-z_\\-0-9\\u{00a0}-\\u{10ffff}])*)", "u")] = typeMatch("class");
+	selectors[regexpu("#((?:\\\\.|[A-Za-z_\\-\\u{00a0}-\\u{10ffff}])(?:\\\\.|[A-Za-z_\\-0-9\\u{00a0}-\\u{10ffff}])*)", "u")] = typeMatch("id");
+	var selectorsSecondHalf = {
 		":(not|matches|has|local|global)\\((\\s*)": nestedPseudoClassStartMatch,
 		":((?:\\\\.|[A-Za-z_\\-0-9])+)\\(": pseudoClassStartMatch,
 		":((?:\\\\.|[A-Za-z_\\-0-9])+)": typeMatch("pseudo-class"),
@@ -185,7 +194,18 @@ var parser = new Parser({
 		"^\\s+": irrelevantSpacingStartMatch,
 		"\\s+": spacingMatch,
 		".": invalidMatch
-	},
+	};
+	var selector;
+	for (selector in selectorsSecondHalf) {
+		if (Object.prototype.hasOwnProperty.call(selectorsSecondHalf, selector)) {
+			selectors[selector] = selectorsSecondHalf[selector];
+		}
+	}
+	return selectors;
+}
+
+var parser = new Parser({
+	selector: getSelectors(),
 	inBrackets: {
 		"/\\*[\\s\\S]*?\\*/": addToCurrent,
 		"\"([^\\\\\"]|\\\\.)*\"": addToCurrent,

lib/stringify.js

@@ -2,11 +2,21 @@
 
 var stringify;
 
-function escape(str) {
+var regexpu = require("regexpu-core");
+var identifierEscapeRegexp = new RegExp(
+	regexpu("(^[^A-Za-z_\\-\\u{00a0}-\\u{10ffff}]|^\\-\\-|[^A-Za-z_0-9\\-\\u{00a0}-\\u{10ffff}])", "ug"),
+	"g"
+);
+
+function escape(str, identifier) {
 	if(str === "*") {
 		return "*";
 	}
-	return str.replace(/(^[^A-Za-z_\\-]|^\-\-|[^A-Za-z_0-9\\-])/g, "\\$1");
+	if (identifier) {
+		return str.replace(identifierEscapeRegexp, "\\$1");
+	} else {
+		return str.replace(/(^[^A-Za-z_\\-]|^\-\-|[^A-Za-z_0-9\\-])/g, "\\$1");
+	}
 }
 
 function stringifyWithoutBeforeAfter(tree) {
@@ -18,9 +28,9 @@ function stringifyWithoutBeforeAfter(tree) {
 	case "element":
 		return (typeof tree.namespace === "string" ? escape(tree.namespace) + "|" : "") + escape(tree.name);
 	case "class":
-		return "." + escape(tree.name);
+		return "." + escape(tree.name, true);
 	case "id":
-		return "#" + escape(tree.name);
+		return "#" + escape(tree.name, true);
 	case "attribute":
 		return "[" + tree.content + "]";
 	case "spacing":

package.json

@@ -32,7 +32,8 @@
   "homepage": "https://github.com/css-modules/css-selector-tokenizer",
   "dependencies": {
     "cssesc": "^0.1.0",
-    "fastparse": "^1.1.1"
+    "fastparse": "^1.1.1",
+    "regexpu-core": "^3.2.0"
   },
   "devDependencies": {
     "chokidar-cli": "^0.2.1",

test/test-cases.js

@@ -65,6 +65,27 @@ module.exports = {
 		])
 	],
 
+	"class name with high BMP character": [
+		".字",
+		singleSelector([
+			{ type: "class", name: "字" }
+		])
+	],
+
+	"class name with emoji": [
+		".🤔",
+		singleSelector([
+			{ type: "class", name: "🤔" }
+		])
+	],
+
+	"class name with multiple emoji": [
+		".👍👌",
+		singleSelector([
+			{ type: "class", name: "👍👌" }
+		])
+	],
+
 	"id name": [
 		"#idName",
 		singleSelector([
@@ -79,6 +100,20 @@ module.exports = {
 		])
 	],
 
+	"id name with latin-1 character": [
+		"#¡",
+		singleSelector([
+			{ type: "id", name: "¡" }
+		])
+	],
+
+	"id name with complex emoji": [
+		".🖖🏼",
+		singleSelector([
+			{ type: "class", name: "🖖🏼" }
+		])
+	],
+
 	"pseudo class": [
 		":before",
 		singleSelector([