From 35f65ae255f915a3e062d8359d655bef9ed341d6 Mon Sep 17 00:00:00 2001
From: =?UTF-8?q?L=C3=A9a=20Saviot?= <lea.saviot@numworks.com>
Date: Thu, 31 Jan 2019 11:55:03 +0100
Subject: [PATCH] [unicode] Handle not properly written UTF8 in nextCodePoint

---
 ion/src/shared/unicode/utf8_decoder.cpp | 8 +++++++-
 1 file changed, 7 insertions(+), 1 deletion(-)

diff --git a/ion/src/shared/unicode/utf8_decoder.cpp b/ion/src/shared/unicode/utf8_decoder.cpp
index 45e92ad49..af0d59d46 100644
--- a/ion/src/shared/unicode/utf8_decoder.cpp
+++ b/ion/src/shared/unicode/utf8_decoder.cpp
@@ -23,7 +23,13 @@ CodePoint UTF8Decoder::nextCodePoint() {
   uint32_t result = last_k_bits(*m_stringPosition++, 8-leadingOnes-1);
   for (int i = 0; i < leadingOnes - 1; i++) {
     result <<= 6;
-    result += (*m_stringPosition++ & 0x3F);
+    char nextChunk = *m_stringPosition++;
+    if (!nextChunk && 0x80) {
+      /* The code point is not properly written. This might be due to a code
+       * point being translated into chars in a too small buffer. */
+      return UCodePointNull;
+    }
+    result += (nextChunk & 0x3F);
   }
   return CodePoint(result);
 }