handle wacky identifier names in templates

src/parse/index.ts

@@ -1,9 +1,11 @@
+import { isIdentifierStart, isIdentifierChar } from 'acorn';
 import { locate, Location } from 'locate-character';
 import fragment from './state/fragment';
 import { whitespace } from '../utils/patterns';
 import { trimStart, trimEnd } from '../utils/trim';
 import getCodeFrame from '../utils/getCodeFrame';
 import reservedNames from '../utils/reservedNames';
+import fullCharCodeAt from '../utils/fullCharCodeAt';
 import hash from './utils/hash';
 import { Node, Parsed } from '../interfaces';
 import CompileError from '../utils/CompileError';
@@ -147,7 +149,22 @@ export class Parser {
 
 	readIdentifier() {
 		const start = this.index;
-		const identifier = this.read(/[a-zA-Z_$][a-zA-Z0-9_$]*/);
+
+		let i = this.index;
+
+		const code = fullCharCodeAt(this.template, i);
+		if (!isIdentifierStart(code, true)) return null;
+
+		i += code <= 0xffff ? 1 : 2;
+
+		while (i < this.template.length) {
+			const code = fullCharCodeAt(this.template, i);
+
+			if (!isIdentifierChar(code, true)) break;
+			i += code <= 0xffff ? 1 : 2;
+		}
+
+		const identifier = this.template.slice(this.index, this.index = i);
 
 		if (reservedNames.has(identifier)) {
 			this.error(`'${identifier}' is a reserved word in JavaScript and cannot be used here`, start);

src/utils/fullCharCodeAt.ts

@@ -0,0 +1,10 @@
+// Adapted from https://github.com/acornjs/acorn/blob/6584815dca7440e00de841d1dad152302fdd7ca5/src/tokenize.js
+// Reproduced under MIT License https://github.com/acornjs/acorn/blob/master/LICENSE
+
+export default function fullCharCodeAt(str: string, i: number): number {
+	let code = str.charCodeAt(i)
+	if (code <= 0xd7ff || code >= 0xe000) return code;
+
+	let next = str.charCodeAt(i + 1);
+	return (code << 10) + next - 0x35fdc00;
+}

src/utils/isValidIdentifier.ts

@@ -1,10 +1,14 @@
 import { isIdentifierStart, isIdentifierChar } from 'acorn';
+import fullCharCodeAt from './fullCharCodeAt';
 
 export default function isValidIdentifier(str: string): boolean {
-	if (!isIdentifierStart(str.charCodeAt(0), true)) return false;
+	let i = 0;
 
-	for (let i = 0; i < str.length; i += 1) {
-		if (!isIdentifierChar(str.charCodeAt(i), true)) return false;
+	while (i < str.length) {
+		const code = fullCharCodeAt(str, i);
+		if (!(i === 0 ? isIdentifierStart : isIdentifierChar)(code, true)) return false;
+
+		i += code <= 0xffff ? 1 : 2;
 	}
 
 	return true;

test/parser/samples/unusual-identifier/input.html

@@ -0,0 +1,3 @@
+{{#each things as 𐊧}}
+	<p>𐊧</p>
+{{/each}}

test/parser/samples/unusual-identifier/output.json

@@ -0,0 +1,41 @@
+{
+	"hash": 2991613308,
+	"html": {
+		"start": 0,
+		"end": 43,
+		"type": "Fragment",
+		"children": [
+			{
+				"start": 0,
+				"end": 43,
+				"type": "EachBlock",
+				"expression": {
+					"type": "Identifier",
+					"start": 8,
+					"end": 14,
+					"name": "things"
+				},
+				"children": [
+					{
+						"start": 24,
+						"end": 33,
+						"type": "Element",
+						"name": "p",
+						"attributes": [],
+						"children": [
+							{
+								"start": 27,
+								"end": 29,
+								"type": "Text",
+								"data": "𐊧"
+							}
+						]
+					}
+				],
+				"context": "𐊧"
+			}
+		]
+	},
+	"css": null,
+	"js": null
+}