Add Win32 Unicode <-> UTF-8 conversion functions.

These will be used by functions that need to use Unicode
Win32 APIS, like GetFullPathNameW() which supports long paths,
instead of GetFullPathNameA() which does not, even  when
long paths are enabled on the system!

Author: digit-google

src/util.cc

@@ -561,7 +561,55 @@ void Win32Fatal(const char* function, const char* hint) {
     Fatal("%s: %s", function, GetLastErrorString().c_str());
   }
 }
-#endif
+
+bool ConvertUTF8ToWin32Unicode(const std::string& input, std::wstring* output,
+                               std::string* err) {
+  output->clear();
+  if (input.empty())
+    return true;
+
+  int int_size = static_cast<int>(input.size());
+  if (static_cast<size_t>(int_size) != input.size()) {
+    *err = "Input string length > INT_MAX";
+    return false;
+  }
+  int wide_size =
+      MultiByteToWideChar(CP_UTF8, 0, input.c_str(), int_size, nullptr, 0);
+  if (wide_size <= 0) {
+    *err = "MultiByteToWideChar(" + input + "): " + GetLastErrorString();
+    return false;
+  }
+  output->resize(static_cast<size_t>(wide_size));
+  MultiByteToWideChar(CP_UTF8, 0, input.c_str(), int_size,
+                      const_cast<wchar_t*>(output->data()), wide_size);
+  return true;
+}
+
+bool ConvertWin32UnicodeToUTF8(const std::wstring& input, std::string* output,
+                               std::string* err) {
+  output->clear();
+  if (input.empty())
+    return true;
+
+  int int_size = static_cast<int>(input.size());
+  if (int_size != input.size()) {
+    *err = "Input string length > INT_MAX";
+    return false;
+  }
+  int utf8_size = WideCharToMultiByte(CP_UTF8, 0, input.c_str(), int_size, NULL,
+                                      0, NULL, NULL);
+  if (utf8_size <= 0) {
+    *err = "WideCharToMultiByte(" + std::string(input.begin(), input.end()) +
+           "): " + GetLastErrorString();
+    return false;
+  }
+
+  output->resize(static_cast<size_t>(utf8_size));
+  WideCharToMultiByte(CP_UTF8, 0, input.c_str(), int_size,
+                      const_cast<char*>(output->data()), utf8_size, NULL, NULL);
+  return true;
+}
+#endif  // _WIN32
 
 bool islatinalpha(int c) {
   // isalpha() is locale-dependent.

src/util.h

@@ -124,6 +124,18 @@ std::string GetLastErrorString();
 /// Calls Fatal() with a function name and GetLastErrorString.
 NORETURN void Win32Fatal(const char* function, const char* hint = NULL);
 
+/// Convert UTF-8 string to Win32 Unicode.
+/// On success, set |*output| then return true.
+/// On Failure, clear |*output|, set |*err| then return false.
+bool ConvertUTF8ToWin32Unicode(const std::string& input, std::wstring* output,
+                               std::string* err);
+
+/// Convert WIN32 Unicode to UTF-8 string.
+/// On success, set |*output| then return true.
+/// On Failure, clear |*output|, set |*err| then return false.
+bool ConvertWin32UnicodeToUTF8(const std::wstring& input, std::string* output,
+                               std::string* err);
+
 /// Naive implementation of C++ 20 std::bit_cast(), used to fix Clang and GCC
 /// [-Wcast-function-type] warning on casting result of GetProcAddress().
 template <class To, class From>

src/util_test.cc

@@ -502,3 +502,25 @@ TEST(StripAnsiEscapeCodes, StripColors) {
   EXPECT_EQ("affixmgr.cxx:286:15: warning: using the result... [-Wparentheses]",
             stripped);
 }
+
+#ifdef _WIN32
+TEST(ConvertWin32UnicodeToUTF8, Test) {
+  std::string output;
+  std::string err;
+  EXPECT_TRUE(ConvertWin32UnicodeToUTF8(std::wstring(L"B\u00E9b\u00E9"),
+                                        &output, &err));
+  EXPECT_TRUE(err.empty()) << err;
+  EXPECT_EQ(output, std::string("B\xC3\xA9"
+                                "b\xC3\xA9"));
+}
+
+TEST(ConvertUTF8ToWin32Unicode, Test) {
+  std::string err;
+  std::wstring output;
+  EXPECT_TRUE(ConvertUTF8ToWin32Unicode(std::string("B\xC3\xA9"
+                                                    "b\xC3\xA9"),
+                                        &output, &err));
+  EXPECT_TRUE(err.empty()) << err;
+  EXPECT_EQ(output, std::wstring(L"B\u00E9b\u00E9"));
+}
+#endif  // _WIN32