From 413c48bc61e8efda8dad3968b26cedad363302ef Mon Sep 17 00:00:00 2001
From: =?UTF-8?q?Felix=20B=C3=B6hm?= <188768+fb55@users.noreply.github.com>
Date: Thu, 7 Apr 2022 10:24:49 +0100
Subject: [PATCH] fix(encode): Handle edge-case with surrogate pairs

---
 src/encode-trie.ts | 42 ++++++++++++++++++++++--------------------
 src/encode.spec.ts |  3 +++
 2 files changed, 25 insertions(+), 20 deletions(-)

diff --git a/src/encode-trie.ts b/src/encode-trie.ts
index e95b24cd..ba0ea83f 100644
--- a/src/encode-trie.ts
+++ b/src/encode-trie.ts
@@ -30,36 +30,38 @@ export function encodeHTMLTrieRe(regExp: RegExp, str: string): string {
 
     while ((match = regExp.exec(str)) !== null) {
         const i = match.index;
+        ret += str.substring(lastIdx, i);
         const char = str.charCodeAt(i);
         let next = htmlTrie.get(char);
 
-        if (next != null) {
-            if (typeof next !== "string") {
-                // We are in a branch. Try to match the next char.
-                if (i + 1 < str.length) {
-                    const value =
-                        typeof next.n === "number"
-                            ? next.n === str.charCodeAt(i + 1)
-                                ? next.o
-                                : undefined
-                            : next.n.get(str.charCodeAt(i + 1));
+        if (typeof next === "object") {
+            // We are in a branch. Try to match the next char.
+            if (i + 1 < str.length) {
+                const nextChar = str.charCodeAt(i + 1);
+                const value =
+                    typeof next.n === "number"
+                        ? next.n === nextChar
+                            ? next.o
+                            : undefined
+                        : next.n.get(nextChar);
 
-                    if (value !== undefined) {
-                        ret += str.substring(lastIdx, i) + value;
-                        lastIdx = regExp.lastIndex += 1;
-                        continue;
-                    }
+                if (value !== undefined) {
+                    ret += value;
+                    lastIdx = regExp.lastIndex += 1;
+                    continue;
                 }
-
-                // If we have a character without a value, use a numeric entitiy.
-                next = next.v ?? `&#x${char.toString(16)};`;
             }
 
-            ret += str.substring(lastIdx, i) + next;
+            next = next.v;
+        }
+
+        // We might have a tree node without a value; skip and use a numeric entitiy.
+        if (next !== undefined) {
+            ret += next;
             lastIdx = i + 1;
         } else {
             const cp = getCodePoint(str, i);
-            ret += `${str.substring(lastIdx, i)}&#x${cp.toString(16)};`;
+            ret += `&#x${cp.toString(16)};`;
             // Increase by 1 if we have a surrogate pair
             lastIdx = regExp.lastIndex += Number(cp !== char);
         }
diff --git a/src/encode.spec.ts b/src/encode.spec.ts
index a0e13432..dd334a98 100644
--- a/src/encode.spec.ts
+++ b/src/encode.spec.ts
@@ -54,6 +54,9 @@ describe("Encode->decode test", () => {
 
     it("should encode trailing parts of entities", () =>
         expect(entities.encodeHTML("\ud835")).toBe("&#xd835;"));
+
+    it("should encode surrogate pair with first surrogate equivalent of entity, without corresponding entity", () =>
+        expect(entities.encodeHTML("\u{1d4a4}")).toBe("&#x1d4a4;"));
 });
 
 describe("encodeNonAsciiHTML", () => {