Merge pull request #21682 from calixteman/fix/xml-invalid-char-ref

Don't throw on an invalid XML character reference

src/core/xml_parser.js

@@ -53,12 +53,12 @@ class XMLParserBase {
   }
 
   _resolveEntities(s) {
-    return s.replaceAll(XMLParserBase._entityRegex, (_, hex, dec, entity) => {
+    return s.replaceAll(XMLParserBase._entityRegex, (all, hex, dec, entity) => {
-      if (hex) {
+      if (hex || dec) {
-        return String.fromCodePoint(parseInt(hex, 16));
+        const code = hex ? parseInt(hex, 16) : parseInt(dec, 10);
-      }
+        // An out-of-range or unparsable code point is kept as-is, since
-      if (dec) {
+        // `String.fromCodePoint` would throw on it.
-        return String.fromCodePoint(parseInt(dec, 10));
+        return code >= 0 && code <= 0x10ffff ? String.fromCodePoint(code) : all;
       }
       switch (entity) {
         case "lt":

test/unit/xml_spec.js

@@ -109,6 +109,27 @@ describe("XML", function () {
     });
   });
 
+  describe("character references", function () {
+    const parseText = xml =>
+      new SimpleXMLParser({}).parseFromString(xml).documentElement.textContent;
+
+    it("should resolve the valid ones", function () {
+      expect(parseText("<a>&#65;&#x42;&#x1F602;&#0;&#x10FFFF;</a>")).toEqual(
+        "AB\u{1F602}\0\u{10FFFF}"
+      );
+    });
+
+    it("should keep the invalid ones as-is", function () {
+      // These must not throw: `String.fromCodePoint` rejects anything which
+      // isn't a code point.
+      expect(parseText("<a>&#xZZ;</a>")).toEqual("&#xZZ;");
+      expect(parseText("<a>&#zz;</a>")).toEqual("&#zz;");
+      expect(parseText("<a>&#x110000;</a>")).toEqual("&#x110000;");
+      expect(parseText("<a>&#1114112;</a>")).toEqual("&#1114112;");
+      expect(parseText("<a>&#-1;</a>")).toEqual("&#-1;");
+    });
+  });
+
   it("should parse processing instructions", function () {
     const xml = `
       <a>