Comment-only update

[wxWidgets.git] / include / wx / unichar.h

diff --git a/include/wx/unichar.h b/include/wx/unichar.h
index 8917f569f6d0ae3b2ae14644480d47dcec2496f8..a661e5a4e26440b23335e40fedec07208df2a305 100644 (file)
--- a/include/wx/unichar.h
+++ b/include/wx/unichar.h
@@ -3,7 +3,6 @@
 // Purpose:     wxUniChar and wxUniCharRef classes
 // Author:      Vaclav Slavik
 // Created:     2007-03-19
 // Purpose:     wxUniChar and wxUniCharRef classes
 // Author:      Vaclav Slavik
 // Created:     2007-03-19

-// RCS-ID:      $Id$

 // Copyright:   (c) 2007 REA Elektronik GmbH
 // Licence:     wxWindows licence
 ///////////////////////////////////////////////////////////////////////////////
 // Copyright:   (c) 2007 REA Elektronik GmbH
 // Licence:     wxWindows licence
 ///////////////////////////////////////////////////////////////////////////////


@@ -13,8 +12,10 @@
 
 #include "wx/defs.h"
 #include "wx/chartype.h"
 
 #include "wx/defs.h"
 #include "wx/chartype.h"

+#include "wx/stringimpl.h"

 
 

-class WXDLLIMPEXP_BASE wxUniCharRef;
+class WXDLLIMPEXP_FWD_BASE wxUniCharRef;
+class WXDLLIMPEXP_FWD_BASE wxString;

 
 // This class represents single Unicode character. It can be converted to
 // and from char or wchar_t and implements commonly used character operations.
 
 // This class represents single Unicode character. It can be converted to
 // and from char or wchar_t and implements commonly used character operations.


@@ -33,30 +34,63 @@ public:
     wxUniChar(char c) { m_value = From8bit(c); }
     wxUniChar(unsigned char c) { m_value = From8bit((char)c); }
 
     wxUniChar(char c) { m_value = From8bit(c); }
     wxUniChar(unsigned char c) { m_value = From8bit((char)c); }
 

-    // Create the character from a wchar_t character value.
-    wxUniChar(wchar_t c) { m_value = c; }
-
-#ifndef wxWINT_T_IS_TYPEDEF
-    // Create the character from a wint_t character value.
-    wxUniChar(wint_t c) { m_value = c; }
-#endif
-
-    wxUniChar(int c) { m_value = c; }
+#define wxUNICHAR_DEFINE_CTOR(type) \
+    wxUniChar(type c) { m_value = (value_type)c; }
+    wxDO_FOR_INT_TYPES(wxUNICHAR_DEFINE_CTOR)
+#undef wxUNICHAR_DEFINE_CTOR

 
     wxUniChar(const wxUniCharRef& c);
 
     // Returns Unicode code point value of the character
     value_type GetValue() const { return m_value; }
 
 
     wxUniChar(const wxUniCharRef& c);
 
     // Returns Unicode code point value of the character
     value_type GetValue() const { return m_value; }
 

-    // Casts to char and wchar_t types:
+#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; }
+
+    // Returns true if the character is representable as a single byte in the
+    // current locale encoding and return this byte in output argument c (which
+    // must be non-NULL)
+    bool GetAsChar(char *c) const
+    {
+#if wxUSE_UNICODE
+        if ( !IsAscii() )
+        {
+#if !wxUSE_UTF8_LOCALE_ONLY
+            if ( GetAsHi8bit(m_value, c) )
+                return true;
+#endif // !wxUSE_UTF8_LOCALE_ONLY
+
+            return false;
+        }
+#endif // wxUSE_UNICODE
+
+        *c = wx_truncate_cast(char, m_value);
+        return true;
+    }
+
+    // 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 unsigned char() const { return (unsigned char)To8bit(m_value); }
     operator char() const { return To8bit(m_value); }
     operator unsigned char() const { return (unsigned char)To8bit(m_value); }

-    operator wchar_t() const { return m_value; }
-#ifndef wxWINT_T_IS_TYPEDEF
-    operator wint_t() const { return m_value; }
-#endif
-    operator int() const { return m_value; }
-    operator unsigned int() const { return m_value; }
+
+#define wxUNICHAR_DEFINE_OPERATOR_PAREN(type) \
+    operator type() const { return (type)m_value; }
+    wxDO_FOR_INT_TYPES(wxUNICHAR_DEFINE_OPERATOR_PAREN)
+#undef wxUNICHAR_DEFINE_OPERATOR_PAREN

 
     // We need this operator for the "*p" part of expressions like "for (
     // const_iterator p = begin() + nStart; *p; ++p )". In this case,
 
     // We need this operator for the "*p" part of expressions like "for (
     // const_iterator p = begin() + nStart; *p; ++p )". In this case,


@@ -65,53 +99,76 @@ public:
     // operator unspecified_bool_type() would only makes the ambiguity worse.)
     operator bool() const { return m_value != 0; }
     bool operator!() const { return !((bool)*this); }
     // operator unspecified_bool_type() would only makes the ambiguity worse.)
     operator bool() const { return m_value != 0; }
     bool operator!() const { return !((bool)*this); }

-#if (defined(__VISUALC__) && __VISUALC__ < 1400) || \
-    defined(__DIGITALMARS__) || defined(__BORLANDC__)
-    // We need this for VC++ < 8 or DigitalMars and expressions like
-    // "str[0] && *p":
+
+    // And this one is needed by some (not all, but not using ifdefs makes the
+    // code easier) compilers to parse "str[0] && *p" successfully

     bool operator&&(bool v) const { return (bool)*this && v; }
     bool operator&&(bool v) const { return (bool)*this && v; }

-#endif

 
     // Assignment operators:
 
     // Assignment operators:

-    wxUniChar& operator=(const wxUniChar& c) { m_value = c.m_value; return *this; }
+    wxUniChar& operator=(const wxUniChar& c) { if (&c != this) m_value = c.m_value; return *this; }
+    wxUniChar& operator=(const wxUniCharRef& c);

     wxUniChar& operator=(char c) { m_value = From8bit(c); return *this; }
     wxUniChar& operator=(char c) { m_value = From8bit(c); return *this; }

-    wxUniChar& operator=(wchar_t c) { m_value = c; return *this; }
-#ifndef wxWINT_T_IS_TYPEDEF
-    wxUniChar& operator=(wint_t c) { m_value = c; return *this; }
-#endif
+    wxUniChar& operator=(unsigned char c) { m_value = From8bit((char)c); return *this; }
+
+#define wxUNICHAR_DEFINE_OPERATOR_EQUAL(type) \
+    wxUniChar& operator=(type c) { m_value = (value_type)c; return *this; }
+    wxDO_FOR_INT_TYPES(wxUNICHAR_DEFINE_OPERATOR_EQUAL)
+#undef wxUNICHAR_DEFINE_OPERATOR_EQUAL

 
     // Comparison operators:
 
     // Comparison operators:

+#define wxDEFINE_UNICHAR_CMP_WITH_INT(T, op) \
+    bool operator op(T c) const { return m_value op (value_type)c; }

 
     // define the given comparison operator for all the types
 
     // define the given comparison operator for all the types

-#define wxDEFINE_UNICHAR_OPERATOR_NO_WINT(op)                                 \
+#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(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(wchar_t c) const { return m_value op (value_type)c; }
-
-#ifdef wxWINT_T_IS_TYPEDEF
-    #define wxDEFINE_UNICHAR_OPERATOR wxDEFINE_UNICHAR_OPERATOR_NO_WINT
-#else // wint_t is a separate type, need to overload for it too
-    #define wxDEFINE_UNICHAR_OPERATOR(op)                                     \
-        wxDEFINE_UNICHAR_OPERATOR_NO_WINT(op)                                 \
-        bool operator op(wint_t c) const { return m_value op (value_type)c; }
-#endif
+    bool operator op(unsigned char c) const { return m_value op From8bit((char)c); } \
+    wxDO_FOR_INT_TYPES_1(wxDEFINE_UNICHAR_CMP_WITH_INT, op)

 
     wxFOR_ALL_COMPARISONS(wxDEFINE_UNICHAR_OPERATOR)
 
 
     wxFOR_ALL_COMPARISONS(wxDEFINE_UNICHAR_OPERATOR)
 

-#undef wxDEFINE_UNICHAR_OPERATOR_NO_WINT

 #undef wxDEFINE_UNICHAR_OPERATOR
 #undef wxDEFINE_UNICHAR_OPERATOR

+#undef wxDEFINE_UNCHAR_CMP_WITH_INT

 
     // 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); }
 
     // 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; }
     int operator-(wchar_t c) const { return m_value - (value_type)c; }

-#ifndef wxWINT_T_IS_TYPEDEF
-    int operator-(wint_t c) const { return m_value - (value_type)c; }
-#endif
+

 
 private:
 
 private:

-    static value_type From8bit(char c);
-    static char To8bit(value_type c);
+    // notice that we implement these functions inline for 7-bit ASCII
+    // characters purely for performance reasons
+    static value_type From8bit(char c)
+    {
+#if wxUSE_UNICODE
+        if ( (unsigned char)c < 0x80 )
+            return c;
+
+        return FromHi8bit(c);
+#else
+        return c;
+#endif
+    }
+
+    static char To8bit(value_type c)
+    {
+#if wxUSE_UNICODE
+        if ( c < 0x80 )
+            return wx_truncate_cast(char, c);
+
+        return ToHi8bit(c);
+#else
+        return c;
+#endif
+    }
+
+    // helpers of the functions above called to deal with non-ASCII chars
+    static value_type FromHi8bit(char c);
+    static char ToHi8bit(value_type v);
+    static bool GetAsHi8bit(value_type v, char *c);

 
 private:
     value_type m_value;
 
 private:
     value_type m_value;


@@ -125,9 +182,14 @@ private:
 class WXDLLIMPEXP_BASE wxUniCharRef
 {
 private:
 class WXDLLIMPEXP_BASE wxUniCharRef
 {
 private:

+    typedef wxStringImpl::iterator iterator;
+

     // create the reference
     // create the reference

-    // FIXME-UTF8: the interface will need changes for UTF-8 build
-    wxUniCharRef(wxChar *pos) : m_pos(pos) {}
+#if wxUSE_UNICODE_UTF8
+    wxUniCharRef(wxString& str, iterator pos) : m_str(str), 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
 
 public:
     // NB: we have to make this public, because we don't have wxString


@@ -135,83 +197,85 @@ 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!):
     //     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!):

-    //
-    // FIXME-UTF8: the interface will need changes for UTF-8 build
-    static wxUniCharRef CreateForString(wxChar *pos)
+#if wxUSE_UNICODE_UTF8
+    static wxUniCharRef CreateForString(wxString& str, iterator pos)
+        { return wxUniCharRef(str, pos); }
+#else
+    static wxUniCharRef CreateForString(iterator pos)

         { return wxUniCharRef(pos); }
         { return wxUniCharRef(pos); }

+#endif

 
     wxUniChar::value_type GetValue() const { return UniChar().GetValue(); }
 
 
     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(); }
+    bool GetAsChar(char *c) const { return UniChar().GetAsChar(c); }
+

     // Assignment operators:
     // 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)
+        { if (&c != this) *this = c.UniChar(); return *this; }

 
 

-    wxUniCharRef& operator=(char c) { return *this = wxUniChar(c); }
-    wxUniCharRef& operator=(wchar_t c) { return *this = wxUniChar(c); }
+#define wxUNICHAR_REF_DEFINE_OPERATOR_EQUAL(type) \
+    wxUniCharRef& operator=(type c) { return *this = wxUniChar(c); }
+    wxDO_FOR_CHAR_INT_TYPES(wxUNICHAR_REF_DEFINE_OPERATOR_EQUAL)
+#undef wxUNICHAR_REF_DEFINE_OPERATOR_EQUAL

 
 

-    // Casts to wxUniChar type:
-    operator char() const { return UniChar(); }
-    operator unsigned char() const { return UniChar(); }
-    operator wchar_t() const { return UniChar(); }
-#ifndef wxWINT_T_IS_TYPEDEF
-    operator wint_t() const { return UniChar(); }
-#endif
-    operator int() const { return UniChar(); }
-    operator unsigned int() const { return UniChar(); }
+    // Conversions to the same types as wxUniChar is convertible too:
+#define wxUNICHAR_REF_DEFINE_OPERATOR_PAREN(type) \
+    operator type() const { return UniChar(); }
+    wxDO_FOR_CHAR_INT_TYPES(wxUNICHAR_REF_DEFINE_OPERATOR_PAREN)
+#undef wxUNICHAR_REF_DEFINE_OPERATOR_PAREN

 
     // see wxUniChar::operator bool etc. for explanation
     operator bool() const { return (bool)UniChar(); }
     bool operator!() const { return !UniChar(); }
 
     // see wxUniChar::operator bool etc. for explanation
     operator bool() const { return (bool)UniChar(); }
     bool operator!() const { return !UniChar(); }

-#if (defined(__VISUALC__) && __VISUALC__ < 1400) || \
-    defined(__DIGITALMARS__) || defined(__BORLANDC__)

     bool operator&&(bool v) const { return UniChar() && v; }
     bool operator&&(bool v) const { return UniChar() && v; }

-#endif
+
+#define wxDEFINE_UNICHARREF_CMP_WITH_INT(T, op) \
+    bool operator op(T c) const { return UniChar() op c; }

 
     // Comparison operators:
 
     // Comparison operators:

-#define wxDEFINE_UNICHARREF_OPERATOR_NO_WINT(op)                              \
+#define wxDEFINE_UNICHARREF_OPERATOR(op)                                      \

     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(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(wchar_t c) const { return UniChar() op c; }
-
-#ifdef wxWINT_T_IS_TYPEDEF
-    #define wxDEFINE_UNICHARREF_OPERATOR wxDEFINE_UNICHARREF_OPERATOR_NO_WINT
-#else // wint_t is a separate type, need to overload for it too
-    #define wxDEFINE_UNICHARREF_OPERATOR(op)                                  \
-        wxDEFINE_UNICHARREF_OPERATOR_NO_WINT(op)                              \
-        bool operator op(wint_t c) const { return UniChar() op c; }
-#endif
+    wxDO_FOR_CHAR_INT_TYPES_1(wxDEFINE_UNICHARREF_CMP_WITH_INT, op)

 
     wxFOR_ALL_COMPARISONS(wxDEFINE_UNICHARREF_OPERATOR)
 
 
     wxFOR_ALL_COMPARISONS(wxDEFINE_UNICHARREF_OPERATOR)
 

-#undef wxDEFINE_UNICHARREF_OPERATOR_NO_WINT

 #undef wxDEFINE_UNICHARREF_OPERATOR
 #undef wxDEFINE_UNICHARREF_OPERATOR

+#undef wxDEFINE_UNICHARREF_CMP_WITH_INT

 
     // for expressions like c-'A':
     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; }
 
     // for expressions like c-'A':
     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; }
     int operator-(wchar_t c) const { return UniChar() - c; }

-#ifndef wxWINT_T_IS_TYPEDEF
-    int operator-(wint_t c) const { return UniChar() - c; }
-#endif

 
 private:
 
 private:

+#if wxUSE_UNICODE_UTF8
+    wxUniChar UniChar() const;
+#else

     wxUniChar UniChar() const { return *m_pos; }
     wxUniChar UniChar() const { return *m_pos; }

-    friend class WXDLLIMPEXP_BASE wxUniChar;
+#endif
+
+    friend class WXDLLIMPEXP_FWD_BASE wxUniChar;

 
 private:
 
 private:

-    // pointer to the character in string
-    wxChar *m_pos;
+    // reference to the string and pointer to the character in string
+#if wxUSE_UNICODE_UTF8
+    wxString& m_str;
+#endif
+    iterator m_pos;

 };
 
 inline wxUniChar::wxUniChar(const wxUniCharRef& c)
 };
 
 inline wxUniChar::wxUniChar(const wxUniCharRef& c)


@@ -219,32 +283,26 @@ inline wxUniChar::wxUniChar(const wxUniCharRef& c)
     m_value = c.UniChar().m_value;
 }
 
     m_value = c.UniChar().m_value;
 }
 

+inline wxUniChar& wxUniChar::operator=(const wxUniCharRef& c)
+{
+    m_value = c.UniChar().m_value;
+    return *this;
+}
+

 // Comparison operators for the case when wxUniChar(Ref) is the second operand
 // implemented in terms of member comparison functions
 
 // Comparison operators for the case when wxUniChar(Ref) is the second operand
 // implemented in terms of member comparison functions
 

-#define wxCMP_REVERSE(c1, c2, op) c2 op c1
+wxDEFINE_COMPARISONS_BY_REV(char, const wxUniChar&)
+wxDEFINE_COMPARISONS_BY_REV(char, const wxUniCharRef&)

 
 

-wxDEFINE_COMPARISONS(char, const wxUniChar&, wxCMP_REVERSE)
-wxDEFINE_COMPARISONS(char, const wxUniCharRef&, wxCMP_REVERSE)
+wxDEFINE_COMPARISONS_BY_REV(wchar_t, const wxUniChar&)
+wxDEFINE_COMPARISONS_BY_REV(wchar_t, const wxUniCharRef&)

 
 

-wxDEFINE_COMPARISONS(wchar_t, const wxUniChar&, wxCMP_REVERSE)
-wxDEFINE_COMPARISONS(wchar_t, const wxUniCharRef&, wxCMP_REVERSE)
-
-#ifndef wxWINT_T_IS_TYPEDEF
-wxDEFINE_COMPARISONS(wint_t, const wxUniChar&, wxCMP_REVERSE)
-wxDEFINE_COMPARISONS(wint_t, const wxUniCharRef&, wxCMP_REVERSE)
-#endif
-
-wxDEFINE_COMPARISONS(const wxUniChar&, const wxUniCharRef&, wxCMP_REVERSE)
-
-#undef wxCMP_REVERSE
+wxDEFINE_COMPARISONS_BY_REV(const wxUniChar&, const wxUniCharRef&)

 
 // for expressions like c-'A':
 inline int operator-(char c1, const wxUniCharRef& c2) { return -(c2 - c1); }
 
 // for expressions like c-'A':
 inline int operator-(char c1, const wxUniCharRef& c2) { return -(c2 - c1); }

-inline int operator-(wchar_t c1, const wxUniCharRef& c2) { return -(c2 - c1); }
-#ifndef wxWINT_T_IS_TYPEDEF
-inline int operator-(wint_t c1, const wxUniCharRef& c2) { return -(c2 - c1); }
-#endif

 inline int operator-(const wxUniChar& c1, const wxUniCharRef& c2) { return -(c2 - c1); }
 inline int operator-(const wxUniChar& c1, const wxUniCharRef& c2) { return -(c2 - c1); }

+inline int operator-(wchar_t c1, const wxUniCharRef& c2) { return -(c2 - c1); }

 
 #endif /* _WX_UNICHAR_H_ */
 
 #endif /* _WX_UNICHAR_H_ */