diff --git a/projects/mtg/include/WFont.h b/projects/mtg/include/WFont.h
index 46400dad0..de1abff7a 100644
--- a/projects/mtg/include/WFont.h
+++ b/projects/mtg/include/WFont.h
@@ -46,6 +46,8 @@ public:
   virtual void SetTracking(float tracking) = 0;
   // Set Base for the character set to use.
   virtual void SetBase(int base) = 0;
+  // Format text.
+  virtual void FormatText(string &s, vector<string>& output) = 0;
   WFont(int inID) : mFontID(inID) {};
   virtual  ~WFont() {};
 };
@@ -66,6 +68,7 @@ public:
   float	GetStringWidth(const char *s) const {return it->GetStringWidth(s);};
   void SetTracking(float tracking) {it->SetTracking(tracking);};
   void SetBase(int base) {it->SetBase(base);};
+  void FormatText(string &s, vector<string>& output);
 
 private:
   JLBFont * it;
@@ -87,6 +90,7 @@ public:
   virtual float	GetStringWidth(const char *s) const;
   void SetTracking(float tracking) {};
   void SetBase(int base) {};
+  void FormatText(string &s, vector<string>& output) {};
 
   virtual void DrawString(const char *s, float x, float y, int align=JGETEXT_LEFT, float leftOffset = 0, float width = 0);
   virtual int GetCode(const u8 *ch, int *charLength) const = 0;
@@ -131,6 +135,7 @@ public:
   void DrawString(const char *s, float x, float y, int align=JGETEXT_LEFT, float leftOffset = 0, float width = 0);
   int GetCode(const u8 *ch, int *charLength) const;
   int GetMana(const u8 *ch) const;
+  void FormatText(string &s, vector<string>& output);
 };
 
 class WUFont : public WFBFont
@@ -141,6 +146,7 @@ public:
 
   int GetCode(const u8 *ch, int *charLength) const;
   int GetMana(const u8 *ch) const;
+  void FormatText(string &s, vector<string>& output);
 };
 
 #endif
diff --git a/projects/mtg/src/CardPrimitive.cpp b/projects/mtg/src/CardPrimitive.cpp
index 737c15844..2d1545dfd 100644
--- a/projects/mtg/src/CardPrimitive.cpp
+++ b/projects/mtg/src/CardPrimitive.cpp
@@ -64,63 +64,8 @@ const vector<string>& CardPrimitive::formattedText()
             s[found] = '/';
             found = s.find_first_of("{}", found + 1);
         }
-        while (s.length() > 0)
-        {
-            std::string::size_type len = neofont ? 24 : 33;
-            std::string::size_type cut = s.find_first_of("., \t)", 0);
-            if (cut >= len || cut == string::npos)
-            {
-                // Fix for single byte character in some complex language like Chinese
-                u8 * src = (u8 *) s.c_str();
-                if (neofont)
-                {
-                    len = 0;
-                    std::string::size_type limit = 24;
-                    while (*src != 0)
-                    {
-                        if (*src > 0x80)
-                        { // Non-ASCII
-                            if (len + 2 > limit && !(((*src & 0xF0) == 0xA0) && ((*(src + 1) & 0xF0) == 0xA0)))
-                                break;
-                            src += 2;
-                            len += 2;
-                        }
-                        else
-                        { // ASCII
-                            if (*src == '/' && (*(src + 1) & 0xF0) == 0xA0)
-                                limit += 3;
-                            if (len + 1 > limit && (*src == '+' || *src == '-' || *src == '/'))
-                                break;
-                            src += 1;
-                            len += 1;
-                        }
-                    }
-                }
-                ftdText.push_back(s.substr(0, len));
-                if (s.length() > len)
-                    s = s.substr(len, s.length() - len);
-                else
-                    s = "";
-            }
-            else
-            {
-                std::string::size_type newcut = cut;
-                while (newcut < len && newcut != string::npos)
-                {
-                    // neofont use space to separate one line
-                    u8 * src = (u8 *) s.c_str();
-                    if (neofont && *src > 0x80)
-                        break;
-                    cut = newcut;
-                    newcut = s.find_first_of("., \t)", newcut + 1);
-                }
-                ftdText.push_back(s.substr(0, cut + 1));
-                if (s.length() > cut + 1)
-                    s = s.substr(cut + 1, s.length() - cut - 1);
-                else
-                    s = "";
-            }
-        }
+        WFont * mFont = resources.GetWFont(Fonts::MAGIC_FONT);
+        mFont->FormatText(s, ftdText);
     }
     return ftdText;
 }
diff --git a/projects/mtg/src/WFont.cpp b/projects/mtg/src/WFont.cpp
index 5c2e7ea2a..e3aa55872 100644
--- a/projects/mtg/src/WFont.cpp
+++ b/projects/mtg/src/WFont.cpp
@@ -114,6 +114,38 @@ WLBFont::WLBFont(int inFontID, const char *fontname, int lineheight, bool useVid
     it = NEW JLBFont(path.c_str(), lineheight, useVideoRAM);
 }
 
+void WLBFont::FormatText(string &s, vector<string>& output)
+{
+    // The way of CardPrimitive::formattedText() in r2081.
+    std::string::size_type len = 30;
+    while (s.length() > 0)
+    {
+        std::string::size_type cut = s.find_first_of("., \t)", 0);
+        if (cut >= len || cut == string::npos)
+        {
+            output.push_back(s.substr(0,len));
+            if (s.length() > len)
+                s = s.substr(len, s.length() - len);
+            else
+                s = "";
+        }
+        else
+        {
+            std::string::size_type newcut = cut;
+            while (newcut < len && newcut != string::npos)
+            {
+                cut = newcut;
+                newcut = s.find_first_of("., \t)", newcut + 1);
+            }
+            output.push_back(s.substr(0,cut+1));
+            if (s.length() > cut+1)
+                s = s.substr(cut+1,s.length() - cut - 1);
+            else
+                s = "";
+        }
+    }
+}
+
 WFBFont::WFBFont(int inFontID, const char *fontname, int lineheight, bool useVideoRAM) :
     WFont(inFontID)
 {
@@ -242,6 +274,7 @@ int WFBFont::PreCacheChar(const u8 *ch)
     u8 gray;
 
     code = this->GetCode(ch, &charLength);
+    if (doubleWidthChar(ch) && mIndex) code = mIndex[code]; // mGBCode[] stores the final code.
 
     if (mGBCode[mCurr] != -1) for (int i = 0; i < mCacheSize; i++)
         if (mGBCode[i] == code) return i;
@@ -261,7 +294,7 @@ int WFBFont::PreCacheChar(const u8 *ch)
 
     if (doubleWidthChar(ch))
     {
-        if (mIndex) code = mIndex[code];
+        //if (mIndex) code = mIndex[code];
         size = mFontSize;
         src = mExtraFont + code * mBytesPerChar;
         offset = 0;
@@ -629,27 +662,7 @@ int WGBKFont::PreCacheChar(const u8 *ch)
     unsigned int size, offset;
     u8 gray;
 
-#if 0
-    if (*ch > 0xA0 && *(ch + 1) > 0xA0)
-    {
-        // get offset to the proper character bits (GB2312 encoding)
-        code = (((u32)(*ch - 0xA1)) * 0x5E + ((u32)(*(ch + 1) - 0xA1)));
-        charLength = 2;
-    }
-    else if (*ch > 0x80)
-    {
-        // get offset to the character space's bits (GBK encoding)
-        code = 0;
-        charLength = 2;
-    }
-    else
-    {
-        code = ((u32)*ch);
-        charLength = 1;
-    }
-#else
     code = this->GetCode(ch, &charLength);
-#endif
 
     if (mGBCode[mCurr] != -1) for (int i = 0; i < mCacheSize; i++)
         if (mGBCode[i] == code) return i;
@@ -1022,6 +1035,68 @@ int WGBKFont::GetMana(const u8 *ch) const
     return mana;
 }
 
+void WGBKFont::FormatText(string &s, vector<string>& output)
+{
+    while (s.length() > 0)
+    {
+        std::string::size_type len = 24;
+        std::string::size_type cut = s.find_first_of("., \t)", 0);
+        if (cut >= len || cut == string::npos)
+        {
+            // Fix for single byte character in some complex language like Chinese
+            u8 * src = (u8 *) s.c_str();
+            //if (neofont)
+            {
+                len = 0;
+                std::string::size_type limit = 24;
+                while (*src != 0)
+                {
+                    if (*src > 0x80)
+                    { // Non-ASCII
+                        if (len + 2 > limit && !(((*src & 0xF0) == 0xA0) && ((*(src + 1) & 0xF0) == 0xA0)))
+                            break;
+                        src += 2;
+                        len += 2;
+                    }
+                    else
+                    { // ASCII
+                        if (*src == '/' && (*(src + 1) & 0xF0) == 0xA0)
+                            limit += 3;
+                        if (len + 1 > limit && (*src == '+' || *src == '-' || *src == '/'))
+                            break;
+                        src += 1;
+                        len += 1;
+                    }
+                }
+            }
+            output.push_back(s.substr(0, len));
+            if (s.length() > len)
+                s = s.substr(len, s.length() - len);
+            else
+                s = "";
+        }
+        else
+        {
+            std::string::size_type newcut = cut;
+            while (newcut < len && newcut != string::npos)
+            {
+                // neofont use space to separate one line
+                u8 * src = (u8 *) s.c_str();
+                //if (neofont && *src > 0x80)
+                if (*src > 0x80)
+                    break;
+                cut = newcut;
+                newcut = s.find_first_of("., \t)", newcut + 1);
+            }
+            output.push_back(s.substr(0, cut + 1));
+            if (s.length() > cut + 1)
+                s = s.substr(cut + 1, s.length() - cut - 1);
+            else
+                s = "";
+        }
+    }
+}
+
 int WUFont::GetCode(const u8 *ch, int *charLength) const
 {
     int code = 0;
@@ -1035,7 +1110,7 @@ int WUFont::GetCode(const u8 *ch, int *charLength) const
     if ((*ch & 0xF8) == 0xF0)
     { // Four bytes
         *charLength = 4;
-        code = ((*ch * 0x7) << 18) + ((*(ch + 1) & 0x3F) << 12) + ((*(ch + 2) & 0x3F) << 6) + ((*(ch + 3) * 0x3F));
+        code = ((*ch & 0x7) << 18) + ((*(ch + 1) & 0x3F) << 12) + ((*(ch + 2) & 0x3F) << 6) + ((*(ch + 3) * 0x3F));
     }
     else if ((*ch & 0xF0) == 0xE0)
     { // Three bytes
@@ -1045,7 +1120,7 @@ int WUFont::GetCode(const u8 *ch, int *charLength) const
     else if ((*ch & 0xE0) == 0xC0)
     { // Two bytes
         *charLength = 2;
-        code = ((*ch & 0x1F) << 6) + ((*(ch + 2) & 0x3F));
+        code = ((*ch & 0x1F) << 6) + ((*(ch + 1) & 0x3F));
     }
     else
     {
@@ -1100,3 +1175,38 @@ int WUFont::GetMana(const u8 *ch) const
     }
     return -1;
 }
+
+void WUFont::FormatText(string &s, vector<string>& output)
+{
+    std::string::size_type limit = 22; //28
+    string delim("., \t)");
+
+    while (s.length() > 0)
+    {
+        u8 * src = (u8 *) s.c_str();
+        std::string::size_type len = 0;
+        std::string::size_type ctr = 0;
+        std::string::size_type lastcut = 0;
+        u8 ch = 0;
+        while ((ch = src[len]) != 0)
+        {
+            ctr += 2;
+            if      ((ch & 0xF8) == 0xF0) len += 4;
+            else if ((ch & 0xF0) == 0xE0) len += 3;
+            else if ((ch & 0xE0) == 0xC0) len += 2;
+            else // ASCII
+            {
+                ctr--; len += 1;
+                if (delim.find(ch) != string::npos) lastcut = len;
+                if (ctr > limit && lastcut) {len = lastcut; break;}
+                //if (ch == ' ') break; // if we use ' ' to break a line.
+            }
+            if (ctr > limit) break;
+        }
+        output.push_back(s.substr(0, len));
+        if (s.length() > len)
+            s = s.substr(len, s.length() - len);
+        else
+            s = "";
+    }
+}