Some more rework of text encoding conversions

4 files changed, 34 insertions, 19 deletions

diff --git a/src/lib/VSDXContentCollector.cpp b/src/lib/VSDXContentCollector.cpp
index 2cef0f5..703b1c8 100644
--- a/src/lib/VSDXContentCollector.cpp
+++ b/src/lib/VSDXContentCollector.cpp
@@ -845,7 +845,7 @@ void libvisio::VSDXContentCollector::_flushText()
         max = (m_charFormats[charIndex].charCount == 0 && m_textStream.size()) ? m_textStream.size() : max;
         std::vector<unsigned char> tmpBuffer(max);
         memcpy(&tmpBuffer[0], &pTextBuffer[textBufferPosition], max);
-        appendCharacters(text, tmpBuffer, 0);
+        appendCharacters(text, tmpBuffer, m_textFormat);
         textBufferPosition += max;
       }
       WPXPropertyList textProps;
@@ -2111,10 +2111,7 @@ void libvisio::VSDXContentCollector::_convertDataToString(WPXString &result, con
     return;
   std::vector<unsigned char> tmpData(data.size());
   memcpy(&tmpData[0], data.getDataBuffer(), data.size());
-  if (format == VSD_TEXT_ANSI)
-    appendCharacters(result, tmpData, 0);
-  else if (format == VSD_TEXT_UTF16)
-    appendCharacters(result, tmpData);
+  appendCharacters(result, tmpData, format);
 }
 
 void libvisio::VSDXContentCollector::collectName(unsigned id, unsigned level, const WPXBinaryData &name, TextFormat format)
diff --git a/src/lib/VSDXTypes.h b/src/lib/VSDXTypes.h
index 4c9b8d2..401ba71 100644
--- a/src/lib/VSDXTypes.h
+++ b/src/lib/VSDXTypes.h
@@ -176,7 +176,20 @@ struct ForeignData
       data(fd.data) {}
 };
 
-enum TextFormat { VSD_TEXT_ANSI, VSD_TEXT_UTF16 };
+enum TextFormat
+{
+  VSD_TEXT_ANSI = 0,
+  VSD_TEXT_GREEK,
+  VSD_TEXT_TURKISH,
+  VSD_TEXT_VIETNAMESE,
+  VSD_TEXT_HEBREW,
+  VSD_TEXT_ARABIC,
+  VSD_TEXT_BALTIC,
+  VSD_TEXT_RUSSIAN,
+  VSD_TEXT_THAI,
+  VSD_TEXT_CENTRAL_EUROPE,
+  VSD_TEXT_UTF16
+};
 
 class VSDXName
 {
diff --git a/src/lib/libvisio_utils.cpp b/src/lib/libvisio_utils.cpp
index 5ac2e92..9fd0eb9 100644
--- a/src/lib/libvisio_utils.cpp
+++ b/src/lib/libvisio_utils.cpp
@@ -164,8 +164,11 @@ double libvisio::readDouble(WPXInputStream *input)
   return tmpUnion.d;
 }
 
-void libvisio::appendCharacters(WPXString &text, const std::vector<unsigned char> &characters, unsigned short charset)
+void libvisio::appendCharacters(WPXString &text, const std::vector<unsigned char> &characters, TextFormat format)
 {
+  if (format == VSD_TEXT_UTF16)
+    return appendCharacters(text, characters);
+
   static const unsigned short cp874map[] =
   {
     0x20AC, 0x0020, 0x0020, 0x0020, 0x0020, 0x2026, 0x0020, 0x0020,
@@ -365,6 +368,7 @@ void libvisio::appendCharacters(WPXString &text, const std::vector<unsigned char
     0x0111, 0x00F1, 0x0323, 0x00F3, 0x00F4, 0x01A1, 0x00F6, 0x00F7,
     0x00F8, 0x00F9, 0x00FA, 0x00FB, 0x00FC, 0x01B0, 0x20AB, 0x00FF
   };
+
   std::vector<unsigned char>::const_iterator iter = characters.begin();
   while (iter != characters.end())
   {
@@ -383,36 +387,36 @@ void libvisio::appendCharacters(WPXString &text, const std::vector<unsigned char
     }
     else
     {
-      switch (charset)
+      switch (format)
       {
-      case 0: // ANSI
+      case VSD_TEXT_ANSI:
         ucs4Character = cp1252map[*iter++ - 0x80];
         break;
-      case 0xa1: // GREEEK
+      case VSD_TEXT_GREEK:
         ucs4Character = cp1253map[*iter++ - 0x80];
         break;
-      case 0xa2: // TURKISH
+      case VSD_TEXT_TURKISH:
         ucs4Character = cp1254map[*iter++ - 0x80];
         break;
-      case 0xa3: // VIETNAMESE
+      case VSD_TEXT_VIETNAMESE:
         ucs4Character = cp1258map[*iter++ - 0x80];
         break;
-      case 0xb1: // HEBREW
+      case VSD_TEXT_HEBREW:
         ucs4Character = cp1255map[*iter++ - 0x80];
         break;
-      case 0xb2: // ARABIC
+      case VSD_TEXT_ARABIC:
         ucs4Character = cp1256map[*iter++ - 0x80];
         break;
-      case 0xba: // BALTIC
+      case VSD_TEXT_BALTIC:
         ucs4Character = cp1257map[*iter++ - 0x80];
         break;
-      case 0xcc: // RUSSIAN
+      case VSD_TEXT_RUSSIAN:
         ucs4Character = cp1251map[*iter++ - 0x80];
         break;
-      case 0xde: // THAI
+      case VSD_TEXT_THAI:
         ucs4Character = cp874map[*iter++ - 0x80];
         break;
-      case 0xee: // CENTRAL EUROPE
+      case VSD_TEXT_CENTRAL_EUROPE:
         ucs4Character = cp1250map[*iter++ - 0x80];
         break;
       default:
diff --git a/src/lib/libvisio_utils.h b/src/lib/libvisio_utils.h
index 030a46e..912f32b 100644
--- a/src/lib/libvisio_utils.h
+++ b/src/lib/libvisio_utils.h
@@ -34,6 +34,7 @@
 #include <vector>
 #include <stdio.h>
 #include <libwpd/libwpd.h>
+#include "VSDXTypes.h"
 
 #ifdef _MSC_VER
 
@@ -94,7 +95,7 @@ uint64_t readU64(WPXInputStream *input);
 
 double readDouble(WPXInputStream *input);
 
-void appendCharacters(WPXString &text, const std::vector<unsigned char> &characters, unsigned short charset);
+void appendCharacters(WPXString &text, const std::vector<unsigned char> &characters, TextFormat format);
 void appendCharacters(WPXString &text, const std::vector<unsigned char> &characters);
 
 class EndOfStreamException