CPackWiX: allow and convert UTF-8 sequences in RTF writer

Modules/CPackWIX.cmake

@@ -52,6 +52,7 @@
 #
 #  If CPACK_RESOURCE_FILE_LICENSE has an .txt extension it is implicitly
 #  converted to RTF by the WiX Generator.
+#  The expected encoding of the .txt file is UTF-8.
 #
 #  With CPACK_WIX_LICENSE_RTF you can override the license file used by the
 #  WiX Generator in case CPACK_RESOURCE_FILE_LICENSE is in an unsupported

Source/CPack/WiX/cmWIXRichTextFormatWriter.cxx

@@ -65,7 +65,41 @@ void cmWIXRichTextFormatWriter::AddText(const std::string& text)
           }
         else
           {
-          File << "[NON-ASCII-" << int(c) << "]";
+            if(c <= 0xC0)
+              {
+              EmitInvalidCodepoint(c);
+              }
+            else if(c < 0xE0 && i+1 < text.size())
+              {
+              EmitUnicodeCodepoint(
+                (text[i+1] & 0x3F) |
+                ((c & 0x1F) << 6)
+              );
+              i+= 1;
+              }
+            else if(c < 0xF0 && i+2 < text.size())
+              {
+              EmitUnicodeCodepoint(
+                (text[i+2] & 0x3F) |
+                ((text[i+1] & 0x3F) << 6) |
+                ((c & 0xF) << 12)
+              );
+              i += 2;
+              }
+            else if(c < 0xF8 && i+3 < text.size())
+              {
+              EmitUnicodeCodepoint(
+                (text[i+3] & 0x3F) |
+                ((text[i+2] & 0x3F) << 6) |
+                ((text[i+1] & 0x3F) << 12) |
+                ((c & 0x7) << 18)
+              );
+              i += 3;
+              }
+            else
+              {
+              EmitInvalidCodepoint(c);
+              }
           }
         }
       break;
@@ -82,6 +116,7 @@ void cmWIXRichTextFormatWriter::WriteHeader()
   ControlWord("deflang1031");
 
   WriteFontTable();
+  WriteColorTable();
   WriteGenerator();
 }
 
@@ -99,6 +134,22 @@ void cmWIXRichTextFormatWriter::WriteFontTable()
   EndGroup();
 }
 
+void cmWIXRichTextFormatWriter::WriteColorTable()
+{
+  StartGroup();
+  ControlWord("colortbl ;");
+  ControlWord("red255");
+  ControlWord("green0");
+  ControlWord("blue0;");
+  ControlWord("red0");
+  ControlWord("green255");
+  ControlWord("blue0;");
+  ControlWord("red0");
+  ControlWord("green0");
+  ControlWord("blue255;");
+  EndGroup();
+}
+
 void cmWIXRichTextFormatWriter::WriteGenerator()
 {
   StartGroup();
@@ -135,3 +186,43 @@ void cmWIXRichTextFormatWriter::EndGroup()
 {
   File.put('}');
 }
+
+void cmWIXRichTextFormatWriter::EmitUnicodeCodepoint(int c)
+{
+  // Do not emit byte order mark (BOM)
+  if(c == 0xFEFF)
+    {
+    return;
+    }
+  else if(c <= 0xFFFF)
+    {
+    EmitUnicodeSurrogate(c);
+    }
+  else
+    {
+    c -= 0x10000;
+    EmitUnicodeSurrogate(((c >> 10) & 0x3FF) + 0xD800);
+    EmitUnicodeSurrogate((c & 0x3FF) + 0xDC00);
+    }
+}
+
+void cmWIXRichTextFormatWriter::EmitUnicodeSurrogate(int c)
+{
+  ControlWord("u");
+  if(c <= 32767)
+    {
+    File << c;
+    }
+  else
+    {
+    File << (c - 65536);
+    }
+  File << "?";
+}
+
+void cmWIXRichTextFormatWriter::EmitInvalidCodepoint(int c)
+{
+  ControlWord("cf1 ");
+  File << "[INVALID-BYTE-" << int(c) << "]";
+  ControlWord("cf0 ");
+}

Source/CPack/WiX/cmWIXRichTextFormatWriter.h

@@ -30,6 +30,7 @@ public:
 private:
   void WriteHeader();
   void WriteFontTable();
+  void WriteColorTable();
   void WriteGenerator();
 
   void WriteDocumentPrefix();
@@ -40,6 +41,11 @@ private:
   void StartGroup();
   void EndGroup();
 
+  void EmitUnicodeCodepoint(int c);
+  void EmitUnicodeSurrogate(int c);
+
+  void EmitInvalidCodepoint(int c);
+
   std::ofstream File;
 };
 

Tests/CPackWiXGenerator/CMakeLists.txt

@@ -51,6 +51,8 @@ set(CPACK_PACKAGE_EXECUTABLES
 
 set(CPACK_WIX_PATCH_FILE "${CMAKE_CURRENT_SOURCE_DIR}/patch.xml")
 
+set(CPACK_RESOURCE_FILE_LICENSE "${CMAKE_CURRENT_SOURCE_DIR}/license.txt")
+
 include(CPack)
 
 cpack_add_install_type(Full DISPLAY_NAME "Everything")

Tests/CPackWiXGenerator/license.txt

@@ -0,0 +1,9 @@
+hello world
+merhaba dünya
+ハローワールド
+привет мир
+مرحبا العالم
+你好世界
+
+4-Byte sequences:
+  Perch (Fish) 𩶘 Elevator 𨋢!