X-Git-Url: https://git.saurik.com/wxWidgets.git/blobdiff_plain/dbecee02eda4e75af63746fca276c4df05ed3b94..f71b0c2d56b03b7823e661aacd05cf1a791e5c2f:/include/wx/unichar.h diff --git a/include/wx/unichar.h b/include/wx/unichar.h index 948018d4a8..defc237114 100644 --- a/include/wx/unichar.h +++ b/include/wx/unichar.h @@ -13,8 +13,10 @@ #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 wxStringIteratorNode; // This class represents single Unicode character. It can be converted to // and from char or wchar_t and implements commonly used character operations. @@ -31,29 +33,55 @@ public: // Create the character from 8bit character value encoded in the current // locale's charset. 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. +#if wxWCHAR_T_IS_REAL_TYPE 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; } + wxUniChar(unsigned int c) { m_value = c; } + wxUniChar(long int c) { m_value = c; } + wxUniChar(unsigned long int c) { m_value = c; } + wxUniChar(short int c) { m_value = c; } + wxUniChar(unsigned short int c) { m_value = c; } 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; } + + // 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; } -#ifndef wxWINT_T_IS_TYPEDEF - operator wint_t() const { return m_value; } + operator unsigned char() const { return (unsigned char)To8bit(m_value); } +#if wxWCHAR_T_IS_REAL_TYPE + operator wchar_t() const { return (wchar_t)m_value; } #endif - operator int() const { return m_value; } + operator int() const { return (int)m_value; } + operator unsigned int() const { return (unsigned int)m_value; } + operator long int() const { return (long int)m_value; } + operator unsigned long int() const { return (unsigned long)m_value; } + operator short int() const { return (short int)m_value; } + operator unsigned short int() const { return (unsigned short int)m_value; } // We need this operator for the "*p" part of expressions like "for ( // const_iterator p = begin() + nStart; *p; ++p )". In this case, @@ -62,74 +90,82 @@ 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=(unsigned char c) { m_value = From8bit((char)c); return *this; } +#if wxWCHAR_T_IS_REAL_TYPE 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 - - // Comparision operators: - bool operator==(const wxUniChar& c) const { return m_value == c.m_value; } - bool operator==(char c) const { return m_value == From8bit(c); } - bool operator==(wchar_t c) const { return m_value == (value_type)c; } -#ifndef wxWINT_T_IS_TYPEDEF - bool operator==(wint_t c) const { return m_value == (value_type)c; } #endif + wxUniChar& operator=(int c) { m_value = c; return *this; } + wxUniChar& operator=(unsigned int c) { m_value = c; return *this; } + wxUniChar& operator=(long int c) { m_value = c; return *this; } + wxUniChar& operator=(unsigned long int c) { m_value = c; return *this; } + wxUniChar& operator=(short int c) { m_value = c; return *this; } + wxUniChar& operator=(unsigned short int c) { m_value = c; return *this; } + + // Comparison operators: + + // 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(unsigned char c) const { return m_value op From8bit((char)c); } \ + wxIF_WCHAR_T_TYPE( 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(unsigned int c) const { return m_value op (value_type)c; } \ + bool operator op(short int c) const { return m_value op (value_type)c; } \ + bool operator op(unsigned short int c) const { return m_value op (value_type)c; } \ + bool operator op(long int c) const { return m_value op (value_type)c; } \ + bool operator op(unsigned long int c) const { return m_value op (value_type)c; } + + wxFOR_ALL_COMPARISONS(wxDEFINE_UNICHAR_OPERATOR) + +#undef wxDEFINE_UNICHAR_OPERATOR + + // 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; } - bool operator!=(const wxUniChar& c) const { return m_value != c.m_value; } - bool operator!=(char c) const { return m_value != From8bit(c); } - bool operator!=(wchar_t c) const { return m_value != (value_type)c; } -#ifndef wxWINT_T_IS_TYPEDEF - bool operator!=(wint_t c) const { return m_value != (value_type)c; } -#endif - bool operator>(const wxUniChar& c) const { return m_value > c.m_value; } - bool operator>(char c) const { return m_value > (value_type)c; } - bool operator>(wchar_t c) const { return m_value > (value_type)c; } -#ifndef wxWINT_T_IS_TYPEDEF - bool operator>(wint_t c) const { return m_value > (value_type)c; } -#endif - - bool operator<(const wxUniChar& c) const { return m_value < c.m_value; } - bool operator<(char c) const { return m_value < From8bit(c); } - bool operator<(wchar_t c) const { return m_value < (value_type)c; } -#ifndef wxWINT_T_IS_TYPEDEF - bool operator<(wint_t c) const { return m_value < (value_type)c; } -#endif +private: + // 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; - bool operator>=(const wxUniChar& c) const { return m_value >= c.m_value; } - bool operator>=(char c) const { return m_value >= From8bit(c); } - bool operator>=(wchar_t c) const { return m_value >= (value_type)c; } -#ifndef wxWINT_T_IS_TYPEDEF - bool operator>=(wint_t c) const { return m_value >= (value_type)c; } + return FromHi8bit(c); +#else + return c; #endif + } - bool operator<=(const wxUniChar& c) const { return m_value <= c.m_value; } - bool operator<=(char c) const { return m_value <= From8bit(c); } - bool operator<=(wchar_t c) const { return m_value <= (value_type)c; } -#ifndef wxWINT_T_IS_TYPEDEF - bool operator<=(wint_t c) const { return m_value <= (value_type)c; } -#endif + static char To8bit(value_type c) + { +#if wxUSE_UNICODE + if ( c < 0x80 ) + return 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-(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; } + return ToHi8bit(c); +#else + return c; #endif + } -private: - static value_type From8bit(char c); - static char To8bit(value_type c); + // helpers of the functions above called to deal with non-ASCII chars + static value_type FromHi8bit(char c); + static char ToHi8bit(value_type c); private: value_type m_value; @@ -143,9 +179,14 @@ private: class WXDLLIMPEXP_BASE wxUniCharRef { private: + typedef wxStringImpl::iterator iterator; + // 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(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 @@ -153,110 +194,102 @@ 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!): - // - // FIXME-UTF8: the interface will need changes for UTF-8 build - static wxUniCharRef CreateForString(wxChar *pos) +#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) + { if (&c != this) *this = c.UniChar(); return *this; } wxUniCharRef& operator=(char c) { return *this = wxUniChar(c); } + wxUniCharRef& operator=(unsigned char c) { return *this = wxUniChar(c); } +#if wxWCHAR_T_IS_REAL_TYPE wxUniCharRef& operator=(wchar_t c) { return *this = wxUniChar(c); } +#endif + wxUniCharRef& operator=(int c) { return *this = wxUniChar(c); } + wxUniCharRef& operator=(unsigned int c) { return *this = wxUniChar(c); } + wxUniCharRef& operator=(short int c) { return *this = wxUniChar(c); } + wxUniCharRef& operator=(unsigned short int c) { return *this = wxUniChar(c); } + wxUniCharRef& operator=(long int c) { return *this = wxUniChar(c); } + wxUniCharRef& operator=(unsigned long int c) { return *this = wxUniChar(c); } - // Casts to wxUniChar type: + // Conversions to the same types as wxUniChar is convertible too: operator char() const { return UniChar(); } + operator unsigned char() const { return UniChar(); } +#if wxWCHAR_T_IS_REAL_TYPE 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(); } + operator short int() const { return UniChar(); } + operator unsigned short int() const { return UniChar(); } + operator long int() const { return UniChar(); } + operator unsigned long int() 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; } -#endif - - // Comparision operators: - bool operator==(const wxUniCharRef& c) const { return m_pos == c.m_pos; } - bool operator==(const wxUniChar& c) const { return UniChar() == c; } - bool operator==(char c) const { return UniChar() == c; } - bool operator==(wchar_t c) const { return UniChar() == c; } -#ifndef wxWINT_T_IS_TYPEDEF - bool operator==(wint_t c) const { return UniChar() == c; } -#endif - - bool operator!=(const wxUniCharRef& c) const { return m_pos != c.m_pos; } - bool operator!=(const wxUniChar& c) const { return UniChar() != c; } - bool operator!=(char c) const { return UniChar() != c; } - bool operator!=(wchar_t c) const { return UniChar() != c; } -#ifndef wxWINT_T_IS_TYPEDEF - bool operator!=(wint_t c) const { return UniChar() != c; } -#endif - bool operator>(const wxUniCharRef& c) const { return UniChar() > c.UniChar(); } - bool operator>(const wxUniChar& c) const { return UniChar() > c; } - bool operator>(char c) const { return UniChar() > c; } - bool operator>(wchar_t c) const { return UniChar() > c; } -#ifndef wxWINT_T_IS_TYPEDEF - bool operator>(wint_t c) const { return UniChar() > c; } -#endif + // 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(unsigned char c) const { return UniChar() op c; } \ + wxIF_WCHAR_T_TYPE( bool operator op(wchar_t c) const { return UniChar() op c; } ) \ + bool operator op(int c) const { return UniChar() op c; } \ + bool operator op(unsigned int c) const { return UniChar() op c; } \ + bool operator op(short int c) const { return UniChar() op c; } \ + bool operator op(unsigned short int c) const { return UniChar() op c; } \ + bool operator op(long int c) const { return UniChar() op c; } \ + bool operator op(unsigned long int c) const { return UniChar() op c; } - bool operator<(const wxUniCharRef& c) const { return UniChar() < c.UniChar(); } - bool operator<(const wxUniChar& c) const { return UniChar() < c; } - bool operator<(char c) const { return UniChar() < c; } - bool operator<(wchar_t c) const { return UniChar() < c; } -#ifndef wxWINT_T_IS_TYPEDEF - bool operator<(wint_t c) const { return UniChar() < c; } -#endif + wxFOR_ALL_COMPARISONS(wxDEFINE_UNICHARREF_OPERATOR) - bool operator>=(const wxUniCharRef& c) const { return UniChar() >= c.UniChar(); } - bool operator>=(const wxUniChar& c) const { return UniChar() >= c; } - bool operator>=(char c) const { return UniChar() >= c; } - bool operator>=(wchar_t c) const { return UniChar() >= c; } -#ifndef wxWINT_T_IS_TYPEDEF - bool operator>=(wint_t c) const { return UniChar() >= c; } -#endif - - bool operator<=(const wxUniCharRef& c) const { return UniChar() <= c.UniChar(); } - bool operator<=(const wxUniChar& c) const { return UniChar() <= c; } - bool operator<=(char c) const { return UniChar() <= c; } - bool operator<=(wchar_t c) const { return UniChar() <= c; } -#ifndef wxWINT_T_IS_TYPEDEF - bool operator<=(wint_t c) const { return UniChar() <= c; } -#endif +#undef wxDEFINE_UNICHARREF_OPERATOR // 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; } -#ifndef wxWINT_T_IS_TYPEDEF - 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 - wxChar *m_pos; + // reference to the string and pointer to the character in string +#if wxUSE_UNICODE_UTF8 + wxStringIteratorNode& m_node; +#endif + iterator m_pos; }; inline wxUniChar::wxUniChar(const wxUniCharRef& c) @@ -264,92 +297,30 @@ inline wxUniChar::wxUniChar(const wxUniCharRef& c) m_value = c.UniChar().m_value; } -// Comparision operators for the case when wxUniChar(Ref) is the second operand: -inline bool operator==(char c1, const wxUniChar& c2) { return c2 == c1; } -inline bool operator==(wchar_t c1, const wxUniChar& c2) { return c2 == c1; } -#ifndef wxWINT_T_IS_TYPEDEF -inline bool operator==(wint_t c1, const wxUniChar& c2) { return c2 == c1; } -#endif - -inline bool operator!=(char c1, const wxUniChar& c2) { return c2 != c1; } -inline bool operator!=(wchar_t c1, const wxUniChar& c2) { return c2 != c1; } -#ifndef wxWINT_T_IS_TYPEDEF -inline bool operator!=(wint_t c1, const wxUniChar& c2) { return c2 != c1; } -#endif - -inline bool operator>(char c1, const wxUniChar& c2) { return c2 < c1; } -inline bool operator>(wchar_t c1, const wxUniChar& c2) { return c2 < c1; } -#ifndef wxWINT_T_IS_TYPEDEF -inline bool operator>(wint_t c1, const wxUniChar& c2) { return c2 < c1; } -#endif - -inline bool operator<(char c1, const wxUniChar& c2) { return c2 > c1; } -inline bool operator<(wchar_t c1, const wxUniChar& c2) { return c2 > c1; } -#ifndef wxWINT_T_IS_TYPEDEF -inline bool operator<(wint_t c1, const wxUniChar& c2) { return c2 > c1; } -#endif +inline wxUniChar& wxUniChar::operator=(const wxUniCharRef& c) +{ + m_value = c.UniChar().m_value; + return *this; +} -inline bool operator>=(char c1, const wxUniChar& c2) { return c2 <= c1; } -inline bool operator>=(wchar_t c1, const wxUniChar& c2) { return c2 <= c1; } -#ifndef wxWINT_T_IS_TYPEDEF -inline bool operator>=(wint_t c1, const wxUniChar& c2) { return c2 <= c1; } -#endif +// Comparison operators for the case when wxUniChar(Ref) is the second operand +// implemented in terms of member comparison functions -inline bool operator<=(char c1, const wxUniChar& c2) { return c2 >= c1; } -inline bool operator<=(wchar_t c1, const wxUniChar& c2) { return c2 >= c1; } -#ifndef wxWINT_T_IS_TYPEDEF -inline bool operator<=(wint_t c1, const wxUniChar& c2) { return c2 >= c1; } -#endif +#define wxCMP_REVERSE(c1, c2, op) c2 op c1 +wxDEFINE_COMPARISONS(char, const wxUniChar&, wxCMP_REVERSE) +wxDEFINE_COMPARISONS(char, const wxUniCharRef&, wxCMP_REVERSE) -inline bool operator==(char c1, const wxUniCharRef& c2) { return c2 == c1; } -inline bool operator==(wchar_t c1, const wxUniCharRef& c2) { return c2 == c1; } -#ifndef wxWINT_T_IS_TYPEDEF -inline bool operator==(wint_t c1, const wxUniCharRef& c2) { return c2 == c1; } -#endif -inline bool operator==(const wxUniChar& c1, const wxUniCharRef& c2) { return c2 == c1; } +wxDEFINE_COMPARISONS(wchar_t, const wxUniChar&, wxCMP_REVERSE) +wxDEFINE_COMPARISONS(wchar_t, const wxUniCharRef&, wxCMP_REVERSE) -inline bool operator!=(char c1, const wxUniCharRef& c2) { return c2 != c1; } -inline bool operator!=(wchar_t c1, const wxUniCharRef& c2) { return c2 != c1; } -#ifndef wxWINT_T_IS_TYPEDEF -inline bool operator!=(wint_t c1, const wxUniCharRef& c2) { return c2 != c1; } -#endif -inline bool operator!=(const wxUniChar& c1, const wxUniCharRef& c2) { return c2 != c1; } +wxDEFINE_COMPARISONS(const wxUniChar&, const wxUniCharRef&, wxCMP_REVERSE) -inline bool operator>(char c1, const wxUniCharRef& c2) { return c2 < c1; } -inline bool operator>(wchar_t c1, const wxUniCharRef& c2) { return c2 < c1; } -#ifndef wxWINT_T_IS_TYPEDEF -inline bool operator>(wint_t c1, const wxUniCharRef& c2) { return c2 < c1; } -#endif -inline bool operator>(const wxUniChar& c1, const wxUniCharRef& c2) { return c2 < c1; } - -inline bool operator<(char c1, const wxUniCharRef& c2) { return c2 > c1; } -inline bool operator<(wchar_t c1, const wxUniCharRef& c2) { return c2 > c1; } -#ifndef wxWINT_T_IS_TYPEDEF -inline bool operator<(wint_t c1, const wxUniCharRef& c2) { return c2 > c1; } -#endif -inline bool operator<(const wxUniChar& c1, const wxUniCharRef& c2) { return c2 > c1; } - -inline bool operator>=(char c1, const wxUniCharRef& c2) { return c2 <= c1; } -inline bool operator>=(wchar_t c1, const wxUniCharRef& c2) { return c2 <= c1; } -#ifndef wxWINT_T_IS_TYPEDEF -inline bool operator>=(wint_t c1, const wxUniCharRef& c2) { return c2 <= c1; } -#endif -inline bool operator>=(const wxUniChar& c1, const wxUniCharRef& c2) { return c2 <= c1; } - -inline bool operator<=(char c1, const wxUniCharRef& c2) { return c2 >= c1; } -inline bool operator<=(wchar_t c1, const wxUniCharRef& c2) { return c2 >= c1; } -#ifndef wxWINT_T_IS_TYPEDEF -inline bool operator<=(wint_t c1, const wxUniCharRef& c2) { return c2 >= c1; } -#endif -inline bool operator<=(const wxUniChar& c1, const wxUniCharRef& c2) { return c2 >= c1; } +#undef wxCMP_REVERSE // 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-(wchar_t c1, const wxUniCharRef& c2) { return -(c2 - c1); } #endif /* _WX_UNICHAR_H_ */