X-Git-Url: https://git.saurik.com/wxWidgets.git/blobdiff_plain/171d61321234d05e20cc9d0b5287f9d3fa2c40a8..1663c65500c557d53e94493b7cf59331852c064d:/include/wx/unichar.h

diff --git a/include/wx/unichar.h b/include/wx/unichar.h
index 6581e3e769..ed7c494d37 100644
--- a/include/wx/unichar.h
+++ b/include/wx/unichar.h
@@ -18,7 +18,8 @@
 // wint_t is just a typedef for wchar_t for many old compilers but for modern
 // ones it's a separate type and we must provide a conversion to it to allow
 // passing wxUniChar[Ref] to functions taking wint_t such as iswalnum() &c
-#if (defined(__GNUC__) && !defined(__DARWIN__)) || \
+#if (defined(__GNUC__) && \
+        !defined(__DARWIN__) && !defined(__OS2__) && !defined(__DOS__)) || \
     (defined(__VISUALC__) && defined(_NATIVE_WCHAR_T_DEFINED))
     #define wxWINT_T_IS_SEPARATE_TYPE
 #endif
@@ -31,7 +32,14 @@
     #define wxIF_WINT_T_TYPE(x)
 #endif // wxWINT_T_IS_SEPARATE_TYPE/!wxWINT_T_IS_SEPARATE_TYPE
 
+// wchar_t seems to be defined as unsigned short by all Windows compilers but
+// unsigned int everywhere else
+#ifndef __WIN32__
+    #define wxWCHAR_T_IS_UINT
+#endif
+
 class WXDLLIMPEXP_BASE wxUniCharRef;
+class WXDLLIMPEXP_BASE wxStringIteratorNode;
 
 // This class represents single Unicode character. It can be converted to
 // and from char or wchar_t and implements commonly used character operations.
@@ -57,18 +65,39 @@ public:
 #endif
 
     wxUniChar(int c) { m_value = c; }
+    wxUniChar(long int c) { m_value = c; }
 
     wxUniChar(const wxUniCharRef& c);
 
     // Returns Unicode code point value of the character
     value_type GetValue() const { return m_value; }
 
+#if wxUSE_UNICODE_UTF8
+    // buffer for single UTF-8 character
+    struct Utf8CharBuffer
+    {
+        char data[5];
+        operator const char*() const { return data; }
+    };
+
+    // returns the character encoded as UTF-8
+    // (NB: implemented in stringops.cpp)
+    Utf8CharBuffer AsUTF8() const;
+#endif // wxUSE_UNICODE_UTF8
+
+    // Returns true if the character is an ASCII character:
+    bool IsAscii() const { return m_value < 0x80; }
+
     // Conversions to char and wchar_t types: all of those are needed to be
     // able to pass wxUniChars to verious standard narrow and wide character
     // functions
     operator char() const { return To8bit(m_value); }
     operator wchar_t() const { return m_value; }
     operator int() const { return m_value; }
+    operator long int() const { return m_value; }
+#ifndef wxWCHAR_T_IS_UINT
+    operator unsigned int() const { return m_value; }
+#endif
 
     // More conversions needed for other standard functions: uchar is for VC++
     // _mbxxx() ones (to which toxxx/isxxx() are mapped when _MBCS is defined)
@@ -97,8 +126,10 @@ public:
     // Assignment operators:
     wxUniChar& operator=(const wxUniChar& c) { m_value = c.m_value; return *this; }
     wxUniChar& operator=(char c) { m_value = From8bit(c); return *this; }
+    wxUniChar& operator=(unsigned char c) { m_value = From8bit((char)c); return *this; }
     wxUniChar& operator=(wchar_t c) { m_value = c; return *this; }
     wxUniChar& operator=(int c) { m_value = c; return *this; }
+    wxUniChar& operator=(long int c) { m_value = c; return *this; }
 #ifdef wxWINT_T_IS_SEPARATE_TYPE
     wxUniChar& operator=(wint_t c) { m_value = c; return *this; }
 #endif
@@ -109,7 +140,10 @@ public:
 #define wxDEFINE_UNICHAR_OPERATOR(op)                                         \
     bool operator op(const wxUniChar& c) const { return m_value op c.m_value; }\
     bool operator op(char c) const { return m_value op From8bit(c); }         \
+    bool operator op(unsigned char c) const { return m_value op From8bit((char)c); } \
     bool operator op(wchar_t c) const { return m_value op (value_type)c; }    \
+    bool operator op(int c) const { return m_value op (value_type)c; }        \
+    bool operator op(long int c) const { return m_value op (value_type)c; }   \
     wxIF_WINT_T_TYPE( bool operator op(wint_t c) const { return m_value op (value_type)c; } )
 
     wxFOR_ALL_COMPARISONS(wxDEFINE_UNICHAR_OPERATOR)
@@ -119,6 +153,7 @@ public:
     // this is needed for expressions like 'Z'-c
     int operator-(const wxUniChar& c) const { return m_value - c.m_value; }
     int operator-(char c) const { return m_value - From8bit(c); }
+    int operator-(unsigned char c) const { return m_value - From8bit((char)c); }
     int operator-(wchar_t c) const { return m_value - (value_type)c; }
 #ifdef wxWINT_T_IS_SEPARATE_TYPE
     int operator-(wint_t c) const { return m_value - (value_type)c; }
@@ -143,7 +178,11 @@ private:
     typedef wxStringImpl::iterator iterator;
 
     // create the reference
+#if wxUSE_UNICODE_UTF8
+    wxUniCharRef(wxStringIteratorNode& node, iterator pos) : m_node(node), m_pos(pos) {}
+#else
     wxUniCharRef(iterator pos) : m_pos(pos) {}
+#endif
 
 public:
     // NB: we have to make this public, because we don't have wxString
@@ -151,27 +190,37 @@ public:
     //     as friend; so at least don't use a ctor but a static function
     //     that must be used explicitly (this is more than using 'explicit'
     //     keyword on ctor!):
+#if wxUSE_UNICODE_UTF8
+    static wxUniCharRef CreateForString(wxStringIteratorNode& node, iterator pos)
+        { return wxUniCharRef(node, pos); }
+#else
     static wxUniCharRef CreateForString(iterator pos)
         { return wxUniCharRef(pos); }
+#endif
 
     wxUniChar::value_type GetValue() const { return UniChar().GetValue(); }
 
+#if wxUSE_UNICODE_UTF8
+    wxUniChar::Utf8CharBuffer AsUTF8() const { return UniChar().AsUTF8(); }
+#endif // wxUSE_UNICODE_UTF8
+
+    bool IsAscii() const { return UniChar().IsAscii(); }
+
     // Assignment operators:
-    wxUniCharRef& operator=(const wxUniCharRef& c)
-    {
-        *m_pos = *c.m_pos;
-        return *this;
-    };
+#if wxUSE_UNICODE_UTF8
+    wxUniCharRef& operator=(const wxUniChar& c);
+#else
+    wxUniCharRef& operator=(const wxUniChar& c) { *m_pos = c; return *this; }
+#endif
 
-    wxUniCharRef& operator=(const wxUniChar& c)
-    {
-        *m_pos = c;
-        return *this;
-    };
+    wxUniCharRef& operator=(const wxUniCharRef& c)
+        { return *this = c.UniChar(); }
 
     wxUniCharRef& operator=(char c) { return *this = wxUniChar(c); }
+    wxUniCharRef& operator=(unsigned char c) { return *this = wxUniChar(c); }
     wxUniCharRef& operator=(wchar_t c) { return *this = wxUniChar(c); }
     wxUniCharRef& operator=(int c) { return *this = wxUniChar(c); }
+    wxUniCharRef& operator=(long int c) { return *this = wxUniChar(c); }
 #ifdef wxWINT_T_IS_SEPARATE_TYPE
     wxUniCharRef& operator=(wint_t c) { return *this = wxUniChar(c); }
 #endif
@@ -179,11 +228,15 @@ public:
     // Conversions to the same types as wxUniChar is convertible too:
     operator char() const { return UniChar(); }
     operator int() const { return UniChar(); }
+    operator long int() const { return UniChar(); }
     operator unsigned char() const { return UniChar(); }
     operator wchar_t() const { return UniChar(); }
 #ifdef wxWINT_T_IS_SEPARATE_TYPE
     operator wint_t() const { return UniChar(); }
 #endif
+#ifndef wxWCHAR_T_IS_UINT
+    operator unsigned int() const { return UniChar(); }
+#endif
 
     // see wxUniChar::operator bool etc. for explanation
     operator bool() const { return (bool)UniChar(); }
@@ -198,7 +251,10 @@ public:
     bool operator op(const wxUniCharRef& c) const { return UniChar() op c.UniChar(); }\
     bool operator op(const wxUniChar& c) const { return UniChar() op c; }     \
     bool operator op(char c) const { return UniChar() op c; }                 \
+    bool operator op(unsigned char c) const { return UniChar() op c; }        \
     bool operator op(wchar_t c) const { return UniChar() op c; }              \
+    bool operator op(int c) const { return UniChar() op c; }                  \
+    bool operator op(long int c) const { return UniChar() op c; }             \
     wxIF_WINT_T_TYPE( bool operator op(wint_t c) const { return UniChar() op c; } )
 
     wxFOR_ALL_COMPARISONS(wxDEFINE_UNICHARREF_OPERATOR)
@@ -209,17 +265,26 @@ public:
     int operator-(const wxUniCharRef& c) const { return UniChar() - c.UniChar(); }
     int operator-(const wxUniChar& c) const { return UniChar() - c; }
     int operator-(char c) const { return UniChar() - c; }
+    int operator-(unsigned char c) const { return UniChar() - c; }
     int operator-(wchar_t c) const { return UniChar() - c; }
 #ifdef wxWINT_T_IS_SEPARATE_TYPE
     int operator-(wint_t c) const { return UniChar() - c; }
 #endif
 
 private:
+#if wxUSE_UNICODE_UTF8
+    wxUniChar UniChar() const;
+#else
     wxUniChar UniChar() const { return *m_pos; }
+#endif
+
     friend class WXDLLIMPEXP_BASE wxUniChar;
 
 private:
-    // pointer to the character in string
+    // reference to the string and pointer to the character in string
+#if wxUSE_UNICODE_UTF8
+    wxStringIteratorNode& m_node;
+#endif
     iterator m_pos;
 };