X-Git-Url: https://git.saurik.com/wxWidgets.git/blobdiff_plain/d1b7ed679221c142e7305498e9cb39279a09500e..5d9b5db01e6b36835af126413461c897026e438a:/include/wx/unichar.h diff --git a/include/wx/unichar.h b/include/wx/unichar.h index 75a01f62db..a661e5a4e2 100644 --- 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 -// RCS-ID: $Id$ // Copyright: (c) 2007 REA Elektronik GmbH // Licence: wxWindows licence /////////////////////////////////////////////////////////////////////////////// @@ -11,9 +10,12 @@ #ifndef _WX_UNICHAR_H_ #define _WX_UNICHAR_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. @@ -23,36 +25,72 @@ public: // NB: this is not wchar_t on purpose, it needs to represent the entire // Unicode code points range and wchar_t may be too small for that // (e.g. on Win32 where wchar_t* is encoded in UTF-16) - typedef unsigned int value_type; + typedef wxUint32 value_type; wxUniChar() : m_value(0) {} // 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. - 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; } - // 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 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 char() const { return (unsigned char)To8bit(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, @@ -61,74 +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); } -#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; } -#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=(unsigned char c) { m_value = From8bit((char)c); return *this; } - 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 +#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 - 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 + // Comparison operators: +#define wxDEFINE_UNICHAR_CMP_WITH_INT(T, op) \ + bool operator op(T c) const { return m_value op (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 + // 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); } \ + wxDO_FOR_INT_TYPES_1(wxDEFINE_UNICHAR_CMP_WITH_INT, op) - 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 + wxFOR_ALL_COMPARISONS(wxDEFINE_UNICHAR_OPERATOR) - 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 +#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; } -#ifndef wxWINT_T_IS_TYPEDEF - 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; @@ -142,9 +182,14 @@ 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(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 @@ -152,110 +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!): - // - // 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); } +#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(); } + bool GetAsChar(char *c) const { return UniChar().GetAsChar(c); } + // 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 wchar_t() const { return UniChar(); } -#ifndef wxWINT_T_IS_TYPEDEF - operator wint_t() const { return UniChar(); } -#endif - operator 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(); } -#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 +#define wxDEFINE_UNICHARREF_CMP_WITH_INT(T, op) \ + bool operator op(T c) const { return UniChar() op c; } - 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 + // 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; } \ + wxDO_FOR_CHAR_INT_TYPES_1(wxDEFINE_UNICHARREF_CMP_WITH_INT, op) - 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 - - 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 +#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; } -#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 + wxString& m_str; +#endif + iterator m_pos; }; inline wxUniChar::wxUniChar(const wxUniCharRef& c) @@ -263,92 +283,26 @@ 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 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 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 wxUniChar& wxUniChar::operator=(const wxUniCharRef& c) +{ + m_value = c.UniChar().m_value; + return *this; +} -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; } +// 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 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_BY_REV(char, const wxUniChar&) +wxDEFINE_COMPARISONS_BY_REV(char, const wxUniCharRef&) -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_BY_REV(wchar_t, const wxUniChar&) +wxDEFINE_COMPARISONS_BY_REV(wchar_t, const wxUniCharRef&) -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_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-(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_ */