From c0cf7b4501efe579f0cbd6c2118c00df2ebc6a13 Mon Sep 17 00:00:00 2001
From: Ai Hoshino <ambiguous404@gmail.com>
Date: Sat, 8 Jul 2023 06:10:49 +0800
Subject: [PATCH] fix decoding invalid UTF-8 input (#3563)

* fix decoding invalid UTF-8 input
Close: https://github.com/oven-sh/bun/issues/3562

* add unittest
---
 src/bun.js/bindings/JSStringDecoder.cpp            | 4 ++--
 test/js/node/string_decoder/string-decoder.test.js | 9 +++++++++
 2 files changed, 11 insertions(+), 2 deletions(-)

diff --git a/src/bun.js/bindings/JSStringDecoder.cpp b/src/bun.js/bindings/JSStringDecoder.cpp
index 5ec2585223..b8c2dd50cd 100644
--- a/src/bun.js/bindings/JSStringDecoder.cpp
+++ b/src/bun.js/bindings/JSStringDecoder.cpp
@@ -129,7 +129,7 @@ uint8_t JSStringDecoder::utf8CheckIncomplete(uint8_t* bufPtr, uint32_t length, u
             m_lastNeed = nb - 1;
         return nb;
     }
-    if (--j < i || nb == -2)
+    if (j == 0 || --j < i || nb == -2)
         return 0;
     nb = utf8CheckByte(bufPtr[j]);
     if (nb >= 0) {
@@ -137,7 +137,7 @@ uint8_t JSStringDecoder::utf8CheckIncomplete(uint8_t* bufPtr, uint32_t length, u
             m_lastNeed = nb - 2;
         return nb;
     }
-    if (--j < i || nb == -2)
+    if (j == 0 || --j < i || nb == -2)
         return 0;
     nb = utf8CheckByte(bufPtr[j]);
     if (nb >= 0) {
diff --git a/test/js/node/string_decoder/string-decoder.test.js b/test/js/node/string_decoder/string-decoder.test.js
index f373266782..aba73401ab 100644
--- a/test/js/node/string_decoder/string-decoder.test.js
+++ b/test/js/node/string_decoder/string-decoder.test.js
@@ -241,3 +241,12 @@ for (const StringDecoder of [FakeStringDecoderCall, RealStringDecoder]) {
     });
   });
 }
+
+it("invalid utf-8 input, pr #3562", () => {
+  const decoder = new RealStringDecoder("utf-8");
+  let output = "";
+  output += decoder.write(Buffer.from("B9", "hex"));
+  output += decoder.write(Buffer.from("A9", "hex"));
+  output += decoder.end();
+  expect(output).toStrictEqual("\uFFFD\uFFFD");
+});