Merge 806b89cf63a496933d721fca85d4e5db4d79cc56 into a0f06a70e3da7afa88da9527c43951bca1f7cef2

Fixes: #2066
PiperOrigin-RevId: 955981465
Change-Id: I6b34dc2e634d4b727a50fb3e21e5d8e25c39a13b

googletest/include/gtest/gtest-printers.h

@@ -524,17 +524,14 @@ inline void PrintTo(bool x, ::std::ostream* os) {
 GTEST_API_ void PrintTo(wchar_t wc, ::std::ostream* os);
 
 GTEST_API_ void PrintTo(char32_t c, ::std::ostream* os);
-inline void PrintTo(char16_t c, ::std::ostream* os) {
-  // TODO(b/418738869): Incorrect for values not representing valid codepoints.
-  // Also see https://github.com/google/googletest/issues/4762.
-  PrintTo(static_cast<char32_t>(c), os);
-}
+
+// Overloads for the UTF-8 and UTF-16 code unit types.  A code unit that
+// encodes a code point all by itself is printed with the U+XXXX notation;
+// one that does not (any non-ASCII char8_t, and the UTF-16 surrogates) is
+// printed as a code unit instead, the way wchar_t is.
+GTEST_API_ void PrintTo(char16_t c, ::std::ostream* os);
 #ifdef __cpp_lib_char8_t
-inline void PrintTo(char8_t c, ::std::ostream* os) {
-  // TODO(b/418738869): Incorrect for values not representing valid codepoints.
-  // Also see https://github.com/google/googletest/issues/4762.
-  PrintTo(static_cast<char32_t>(c), os);
-}
+GTEST_API_ void PrintTo(char8_t c, ::std::ostream* os);
 #endif
 
 // gcc/clang __{u,}int128_t
@@ -945,13 +942,13 @@ template <typename T>
 class [[nodiscard]] UniversalPrinter<std::optional<T>> {
  public:
   static void Print(const std::optional<T>& value, ::std::ostream* os) {
-    *os << '(';
     if (!value) {
-      *os << "nullopt";
+      UniversalPrint(std::nullopt, os);
     } else {
+      *os << '(';
       UniversalPrint(*value, os);
+      *os << ')';
     }
-    *os << ')';
   }
 };
 
@@ -961,27 +958,38 @@ class [[nodiscard]] UniversalPrinter<std::nullopt_t> {
   static void Print(std::nullopt_t, ::std::ostream* os) { *os << "(nullopt)"; }
 };
 
+struct UniversalPrinterVisitor {
+  template <typename T>
+  void operator()(const T& arg) const {
+    *os << "'" << GetTypeName<T>() << "(index = " << index << ")' with value ";
+    UniversalPrint(arg, os);
+  }
+  ::std::ostream* os;
+  std::size_t index;
+};
+
 // Printer for std::variant
 template <typename... T>
 class [[nodiscard]] UniversalPrinter<std::variant<T...>> {
  public:
   static void Print(const std::variant<T...>& value, ::std::ostream* os) {
-    *os << '(';
-    std::visit(Visitor{os, value.index()}, value);
-    *os << ')';
-  }
-
- private:
-  struct Visitor {
-    template <typename U>
-    void operator()(const U& u) const {
-      *os << "'" << GetTypeName<U>() << "(index = " << index
-          << ")' with value ";
-      UniversalPrint(u, os);
+    if (value.valueless_by_exception()) {
+      *os << "(valueless)";
+    } else {
+      *os << '(';
+      std::visit(UniversalPrinterVisitor{os, value.index()}, value);
+      *os << ')';
     }
-    ::std::ostream* os;
-    std::size_t index;
-  };
+  }
+};
+
+// Printer for std::monostate
+template <>
+class [[nodiscard]] UniversalPrinter<std::monostate> {
+ public:
+  static void Print(std::monostate, ::std::ostream* os) {
+    *os << "(monostate)";
+  }
 };
 
 // UniversalPrintArray(begin, len, os) prints an array of 'len'

googletest/src/gtest-printers.cc

@@ -285,6 +285,28 @@ void PrintTo(char32_t c, ::std::ostream* os) {
       << static_cast<uint32_t>(c);
 }
 
+// char8_t and char16_t hold code units, not code points, so the U+XXXX
+// notation only applies to the values that encode a code point on their own.
+// The rest -- non-ASCII char8_t, which are always part of a multi-unit UTF-8
+// sequence, and the UTF-16 surrogates -- are printed as code units.
+void PrintTo(char16_t c, ::std::ostream* os) {
+  if (c >= 0xD800 && c <= 0xDFFF) {
+    PrintCharAndCodeTo(c, os);
+  } else {
+    PrintTo(static_cast<char32_t>(c), os);
+  }
+}
+
+#ifdef __cpp_lib_char8_t
+void PrintTo(char8_t c, ::std::ostream* os) {
+  if (c >= 0x80) {
+    PrintCharAndCodeTo(c, os);
+  } else {
+    PrintTo(static_cast<char32_t>(c), os);
+  }
+}
+#endif
+
 // gcc/clang __{u,}int128_t
 #if defined(__SIZEOF_INT128__)
 void PrintTo(__uint128_t v, ::std::ostream* os) {

googletest/test/googletest-printers-test.cc

@@ -396,12 +396,24 @@ TEST(PrintCharTest, UnsignedChar) {
   EXPECT_EQ("'b' (98, 0x62)", Print(static_cast<unsigned char>('b')));
 }
 
-TEST(PrintCharTest, Char16) { EXPECT_EQ("U+0041", Print(u'A')); }
+TEST(PrintCharTest, Char16) {
+  EXPECT_EQ("U+0041", Print(u'A'));
+  EXPECT_EQ("U+754C", Print(u'界'));
+  // Surrogates are not code points, so they are printed as code units.
+  EXPECT_EQ("u'\\xD800' (55296)", Print(static_cast<char16_t>(0xD800)));
+  EXPECT_EQ("u'\\xDFFF' (57343)", Print(static_cast<char16_t>(0xDFFF)));
+}
 
 TEST(PrintCharTest, Char32) { EXPECT_EQ("U+0041", Print(U'A')); }
 
 #ifdef __cpp_lib_char8_t
-TEST(PrintCharTest, Char8) { EXPECT_EQ("U+0041", Print(u8'A')); }
+TEST(PrintCharTest, Char8) {
+  EXPECT_EQ("U+0041", Print(u8'A'));
+  // Only ASCII code units encode a code point on their own; the rest are
+  // printed as code units.
+  EXPECT_EQ("u8'\\x80' (128)", Print(static_cast<char8_t>(0x80)));
+  EXPECT_EQ("u8'\\xFF' (255)", Print(static_cast<char8_t>(0xFF)));
+}
 #endif
 
 // Tests printing other simple, built-in types.
@@ -456,7 +468,7 @@ TEST(PrintBuiltInTypeTest, Integer) {
 #ifdef __cpp_lib_char8_t
   EXPECT_EQ("U+0000",
             Print(std::numeric_limits<char8_t>::min()));  // char8_t
-  EXPECT_EQ("U+00FF",
+  EXPECT_EQ("u8'\\xFF' (255)",
             Print(std::numeric_limits<char8_t>::max()));  // char8_t
 #endif
   EXPECT_EQ("U+0000",
@@ -2030,6 +2042,26 @@ TEST(PrintOneofTest, Basic) {
       PrintToString(Type(NonPrintable{})));
 }
 
+TEST(PrintVariantTest, Monostate) {
+  EXPECT_EQ("(monostate)", PrintToString(std::monostate()));
+
+#if GTEST_HAS_EXCEPTIONS
+  struct ThrowOnMove {
+    ThrowOnMove() = default;
+    ThrowOnMove(ThrowOnMove&& other) { *this = std::move(other); }
+    ThrowOnMove& operator=(ThrowOnMove&&) {
+      (void)std::vector<bool>().at(0);
+      return *this;
+    }
+  };
+  std::variant<std::monostate, ThrowOnMove> v = std::monostate();
+  EXPECT_EQ("('std::monostate(index = 0)' with value (monostate))",
+            PrintToString(v));
+  EXPECT_THROW(v = ThrowOnMove(), std::out_of_range);
+  EXPECT_EQ("(valueless)", PrintToString(v));
+#endif
+}
+
 #if GTEST_INTERNAL_HAS_COMPARE_LIB
 TEST(PrintOrderingTest, Basic) {
   EXPECT_EQ("(less)", PrintToString(std::strong_ordering::less));