X-Git-Url: https://git.saurik.com/wxWidgets.git/blobdiff_plain/153bc803a84c85517d88ae4f7f9d5e28cd54c55d..14619f10b0bdb630206607abd0ce0319d45e095a:/include/wx/unichar.h

diff --git a/include/wx/unichar.h b/include/wx/unichar.h
index b7fe54d9ca..f2ca9c3df4 100644
--- a/include/wx/unichar.h
+++ b/include/wx/unichar.h
@@ -15,29 +15,8 @@
 #include "wx/chartype.h"
 #include "wx/stringimpl.h"
 
-// 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__) && !defined(__OS2__)) || \
-    (defined(__VISUALC__) && defined(_NATIVE_WCHAR_T_DEFINED))
-    #define wxWINT_T_IS_SEPARATE_TYPE
-#endif
-
-// helper macro for doing something dependent on whether wint_t is or isn't a
-// typedef inside another macro
-#ifdef wxWINT_T_IS_SEPARATE_TYPE
-    #define wxIF_WINT_T_TYPE(x) x
-#else // !wxWINT_T_IS_SEPARATE_TYPE
-    #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_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.
@@ -56,38 +35,63 @@ public:
     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; }
-#ifdef wxWINT_T_IS_SEPARATE_TYPE
-    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; }
 
+#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 wchar_t() const { return m_value; }
-    operator 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)
-    // and some wide character functions take wint_t which happens to be the
-    // same as wchar_t for Windows compilers but not for g++ (except for the
-    // special Apple version)
     operator unsigned char() const { return (unsigned char)To8bit(m_value); }
-#ifdef wxWINT_T_IS_SEPARATE_TYPE
-    operator wint_t() const { return m_value; }
-#endif
+
+#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,
@@ -96,46 +100,76 @@ public:
     // operator unspecified_bool_type() would only makes the ambiguity worse.)
     operator bool() const { return m_value != 0; }
     bool operator!() const { return !((bool)*this); }
-#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; }
-#endif
 
     // 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=(wchar_t c) { m_value = c; return *this; }
-    wxUniChar& operator=(int c) { m_value = c; return *this; }
-#ifdef wxWINT_T_IS_SEPARATE_TYPE
-    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:
+#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 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(wchar_t 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; } )
+    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)
 
 #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); }
+    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; }
-#endif
+
 
 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;
@@ -152,7 +186,11 @@ private:
     typedef wxStringImpl::iterator iterator;
 
     // create the reference
+#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
@@ -160,78 +198,84 @@ 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(wxString& str, iterator pos)
+        { return wxUniCharRef(str, pos); }
+#else
     static wxUniCharRef CreateForString(iterator pos)
         { return wxUniCharRef(pos); }
+#endif
 
     wxUniChar::value_type GetValue() const { return UniChar().GetValue(); }
 
-    // Assignment operators:
-    wxUniCharRef& operator=(const wxUniCharRef& c)
-    {
-        *m_pos = *c.m_pos;
-        return *this;
-    };
+#if wxUSE_UNICODE_UTF8
+    wxUniChar::Utf8CharBuffer AsUTF8() const { return UniChar().AsUTF8(); }
+#endif // wxUSE_UNICODE_UTF8
 
-    wxUniCharRef& operator=(const wxUniChar& c)
-    {
-        *m_pos = c;
-        return *this;
-    };
+    bool IsAscii() const { return UniChar().IsAscii(); }
+    bool GetAsChar(char *c) const { return UniChar().GetAsChar(c); }
 
-    wxUniCharRef& operator=(char c) { return *this = wxUniChar(c); }
-    wxUniCharRef& operator=(wchar_t c) { return *this = wxUniChar(c); }
-    wxUniCharRef& operator=(int c) { return *this = wxUniChar(c); }
-#ifdef wxWINT_T_IS_SEPARATE_TYPE
-    wxUniCharRef& operator=(wint_t c) { return *this = wxUniChar(c); }
+    // Assignment operators:
+#if wxUSE_UNICODE_UTF8
+    wxUniCharRef& operator=(const wxUniChar& c);
+#else
+    wxUniCharRef& operator=(const wxUniChar& c) { *m_pos = c; return *this; }
 #endif
 
+    wxUniCharRef& operator=(const wxUniCharRef& c)
+        { if (&c != this) *this = c.UniChar(); return *this; }
+
+#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
+
     // Conversions to the same types as wxUniChar is convertible too:
-    operator char() const { return UniChar(); }
-    operator 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
+#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(); }
-#if (defined(__VISUALC__) && __VISUALC__ < 1400) || \
-    defined(__DIGITALMARS__) || defined(__BORLANDC__)
     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:
 #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(char c) const { return UniChar() op c; }                 \
-    bool operator op(wchar_t c) const { return UniChar() op c; }              \
-    wxIF_WINT_T_TYPE( bool operator op(wint_t c) const { return UniChar() op c; } )
+    wxDO_FOR_CHAR_INT_TYPES_1(wxDEFINE_UNICHARREF_CMP_WITH_INT, op)
 
     wxFOR_ALL_COMPARISONS(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; }
+    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; }
-    friend class WXDLLIMPEXP_BASE wxUniChar;
+#endif
+
+    friend class WXDLLIMPEXP_FWD_BASE wxUniChar;
 
 private:
-    // pointer to the character in string
+    // reference to the string and pointer to the character in string
+#if wxUSE_UNICODE_UTF8
+    wxString& m_str;
+#endif
     iterator m_pos;
 };
 
@@ -240,32 +284,26 @@ inline wxUniChar::wxUniChar(const wxUniCharRef& c)
     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
 
-#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)
-
-#ifdef wxWINT_T_IS_SEPARATE_TYPE
-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); }
 inline int operator-(const wxUniChar& c1, const wxUniCharRef& c2) { return -(c2 - c1); }
 inline int operator-(wchar_t c1, const wxUniCharRef& c2) { return -(c2 - c1); }
-#ifdef wxWINT_T_IS_SEPARATE_TYPE
-inline int operator-(wint_t c1, const wxUniCharRef& c2) { return -(c2 - c1); }
-#endif
 
 #endif /* _WX_UNICHAR_H_ */