From a11e5e021b5c30a0d6490cfff0399a14a3fb798a Mon Sep 17 00:00:00 2001
From: Ben Boeckel <ben.boeckel@kitware.com>
Date: Thu, 14 Mar 2019 11:01:27 -0400
Subject: cm_utf8: reject UTF-16 surrogate half codepoints

---
 Source/cm_utf8.c            | 5 +++++
 Tests/CMakeLib/testUTF8.cxx | 4 ++--
 2 files changed, 7 insertions(+), 2 deletions(-)

diff --git a/Source/cm_utf8.c b/Source/cm_utf8.c
index 52af4a6..2459c02 100644
--- a/Source/cm_utf8.c
+++ b/Source/cm_utf8.c
@@ -71,6 +71,11 @@ const char* cm_utf8_decode_character(const char* first, const char* last,
       return 0;
     }
 
+    /* UTF-16 surrogate halves. */
+    if (0xD800 <= uc && uc <= 0xDFFF) {
+      return 0;
+    }
+
     *pc = uc;
     return first;
   }
diff --git a/Tests/CMakeLib/testUTF8.cxx b/Tests/CMakeLib/testUTF8.cxx
index e225152..f1da6df 100644
--- a/Tests/CMakeLib/testUTF8.cxx
+++ b/Tests/CMakeLib/testUTF8.cxx
@@ -29,8 +29,6 @@ static test_utf8_entry const good_entry[] = {
   { 3, "\xED\x80\x80\x00", 0xD000 },   /* Valid 0xED prefixed codepoint.  */
   { 4, "\xF4\x8F\xBF\xBF", 0x10FFFF }, /* Highest valid RFC codepoint. */
   /* These are invalid according to the RFC, but accepted here. */
-  { 3, "\xED\xA0\x80\x00", 0xD800 },   /* UTF-16 surrogate half. */
-  { 3, "\xED\xBF\xBF\x00", 0xDFFF },   /* UTF-16 surrogate half. */
   { 4, "\xF4\x90\x80\x80", 0x110000 }, /* Lowest out-of-range codepoint. */
   { 4, "\xF5\x80\x80\x80",
     0x140000 }, /* Prefix forces out-of-range codepoints. */
@@ -45,6 +43,8 @@ static test_utf8_char const bad_chars[] = {
   "\xE0\x00\x00\x00", /* Missing continuation bytes. */
   "\xE0\x80\x80\x00", /* Overlong encoding. */
   "\xF0\x80\x80\x80", /* Overlong encoding. */
+  "\xED\xA0\x80\x00", /* UTF-16 surrogate half. */
+  "\xED\xBF\xBF\x00", /* UTF-16 surrogate half. */
   { 0, 0, 0, 0, 0 }
 };
 
-- 
cgit v0.12