Add UnicodeBuffer as first class data type and allow plStrings to be constructed with it

Sources/Plasma/CoreLib/plString.cpp

@@ -204,19 +204,25 @@ void plString::IConvertFromWchar(const wchar_t *wstr, size_t size)
     // We assume that if sizeof(wchar_t) == 2, the data is UTF-16 already
     IConvertFromUtf16((const uint16_t *)wstr, size);
 #else
+    IConvertFromUtf32((const UniChar *)wstr, size);
+#endif
+}
+
+void plString::IConvertFromUtf32(const UniChar *ustr, size_t size)
+{
     fUtf8Buffer = plStringBuffer<char>();
-    if (wstr == nil)
+    if (ustr == nil)
         return;
 
     if ((int32_t)size < 0)
-        size = wcsnlen(wstr, -(int32_t)size);
+        size = ustrlen(ustr, -(int32_t)size);
 
     // Calculate the UTF-8 size
     size_t convlen = 0;
-    const wchar_t *sp = wstr;
+    const UniChar *sp = ustr;
-    while (sp < wstr + size) {
+    while (sp < ustr + size) {
         if (*sp > 0x10FFFF) {
-            hsAssert(0, "UCS-4 character out of range");
+            hsAssert(0, "UTF-32 character out of range");
             convlen += 3;   // Use U+FFFD for release builds
         }
         else if (*sp > 0xFFFF)
@@ -233,8 +239,8 @@ void plString::IConvertFromWchar(const wchar_t *wstr, size_t size)
     // And perform the actual conversion
     char *utf8 = fUtf8Buffer.CreateWritableBuffer(convlen);
     char *dp = utf8;
-    sp = wstr;
+    sp = ustr;
-    while (sp < wstr + size) {
+    while (sp < ustr + size) {
         if (*sp > 0x10FFFF) {
             // Character out of range; Use U+FFFD instead
             *dp++ = 0xE0 | ((BADCHAR_REPLACEMENT >> 12) & 0x0F);
@@ -258,7 +264,6 @@ void plString::IConvertFromWchar(const wchar_t *wstr, size_t size)
         ++sp;
     }
     utf8[convlen] = 0;
-#endif
 }
 
 void plString::IConvertFromIso8859_1(const char *astr, size_t size)
@@ -462,9 +467,9 @@ plStringBuffer<char> plString::ToIso8859_1() const
     return result;
 }
 
-plStringBuffer<UniChar> plString::GetUnicodeArray() const
+plUnicodeBuffer plString::GetUnicodeArray() const
 {
-    plStringBuffer<UniChar> result;
+    plUnicodeBuffer result;
     if (IsEmpty())
         return result;
 
@@ -822,3 +827,11 @@ plStringStream &plStringStream::operator<<(unsigned int num)
     snprintf(buffer, 12, "%u", num);
     return operator<<(buffer);
 }
+
+size_t ustrlen(const UniChar *ustr, size_t max)
+{
+    size_t length = 0;
+    for ( ; *ustr++ && max--; ++length)
+        ;
+    return length;
+}

Sources/Plasma/CoreLib/plString.h

@@ -121,7 +121,6 @@ public:
 
     const _Ch *GetData() const { return IHaveACow() ? fData->fStringData : fShort; }
     size_t GetSize() const { return fSize; }
-
     operator const _Ch *() const { return GetData(); }
 
     // From Haxxia with love
@@ -142,14 +141,15 @@ public:
     }
 };
 
+typedef plStringBuffer<UniChar> plUnicodeBuffer;
+
 
 class plString
 {
+public:
     enum {
         kSizeAuto = (size_t)(0x80000000)
     };
-
-public:
     static const plString Null;
 
 private:
@@ -158,6 +158,7 @@ private:
     void IConvertFromUtf8(const char *utf8, size_t size);
     void IConvertFromUtf16(const uint16_t *utf16, size_t size);
     void IConvertFromWchar(const wchar_t *wstr, size_t size);
+    void IConvertFromUtf32(const UniChar *ustr, size_t size);
     void IConvertFromIso8859_1(const char *astr, size_t size);
 
 public:
@@ -166,10 +167,12 @@ public:
     plString(const char *cstr) { IConvertFromUtf8(cstr, kSizeAuto); }
     plString(const plString &copy) : fUtf8Buffer(copy.fUtf8Buffer) { }
     plString(const plStringBuffer<char> &init) { operator=(init); }
+    plString(const plUnicodeBuffer &init) { IConvertFromUtf32(init.GetData(), init.GetSize()); }
 
     plString &operator=(const char *cstr) { IConvertFromUtf8(cstr, kSizeAuto); return *this; }
     plString &operator=(const plString &copy) { fUtf8Buffer = copy.fUtf8Buffer; return *this; }
     plString &operator=(const plStringBuffer<char> &init);
+    plString &operator=(const plUnicodeBuffer &init) { IConvertFromUtf32(init.GetData(), init.GetSize()); }
 
     plString &operator+=(const char *cstr) { return operator=(*this + cstr); }
     plString &operator+=(const plString &str) { return operator=(*this + str); }
@@ -213,7 +216,7 @@ public:
     plStringBuffer<char> ToIso8859_1() const;
 
     // For use in displaying characters in a GUI
-    plStringBuffer<UniChar> GetUnicodeArray() const;
+    plUnicodeBuffer GetUnicodeArray() const;
 
     size_t GetSize() const { return fUtf8Buffer.GetSize(); }
     bool IsEmpty() const { return fUtf8Buffer.GetSize() == 0; }
@@ -399,6 +402,13 @@ public:
             return ch;
         }
 
+        UniChar operator[](size_t offset) const
+        {
+            iterator copy(*this);
+            copy += offset;
+            return *copy;
+        }
+
         bool AtEnd() const { return m_ptr >= m_end; }
         bool IsValid() const { return m_ptr != 0; }
 
@@ -464,4 +474,6 @@ private:
     size_t fLength;
 };
 
+size_t ustrlen(const UniChar *ustr, size_t max = plString::kSizeAuto);
+
 #endif //plString_Defined