X-Git-Url: https://git.saurik.com/wxWidgets.git/blobdiff_plain/6001e347baf63a91be525f3b9beeefde4bc16f19..eedf3bcbb39a0881f065e124ab3aadbf725dcbca:/include/wx/strconv.h diff --git a/include/wx/strconv.h b/include/wx/strconv.h index 3ff2d7a8db..58b9d75448 100644 --- a/include/wx/strconv.h +++ b/include/wx/strconv.h @@ -6,13 +6,13 @@ // Created: 29/01/98 // RCS-ID: $Id$ // Copyright: (c) 1998 Ove Kaaven, Robert Roebling, Vadim Zeitlin -// Licence: wxWindows license +// Licence: wxWindows licence /////////////////////////////////////////////////////////////////////////////// #ifndef _WX_WXSTRCONVH__ #define _WX_WXSTRCONVH__ -#ifdef __GNUG__ +#if defined(__GNUG__) && !defined(NO_GCC_PRAGMA) #pragma interface "strconv.h" #endif @@ -20,163 +20,331 @@ #include "wx/wxchar.h" #include "wx/buffer.h" +#ifdef __DIGITALMARS__ +#include "typeinfo.h" +#endif + +#if defined(__VISAGECPP__) && __IBMCPP__ >= 400 +# undef __BSEXCPT__ +#endif + #include #if wxUSE_WCHAR_T -//--------------------------------------------------------------------------- -// wxMBConv (base class for conversions, using libc conversion itself) -//--------------------------------------------------------------------------- +// ---------------------------------------------------------------------------- +// wxMBConv (abstract base class for conversions) +// ---------------------------------------------------------------------------- -class WXDLLEXPORT wxMBConv +class WXDLLIMPEXP_BASE wxMBConv { public: - - // the actual conversion takes place here - virtual size_t MB2WC(wchar_t *buf, const char *psz, size_t n) const; - virtual size_t WC2MB(char *buf, const wchar_t *psz, size_t n) const; - - // No longer inline since BC++ complains. - const wxWCharBuffer cMB2WC(const char *psz) const; - const wxCharBuffer cWC2MB(const wchar_t *psz) const; + // the actual conversion takes place here + // + // note that outputSize is the size of the output buffer, not the length of input + // (the latter is always supposed to be NUL-terminated) + virtual size_t MB2WC(wchar_t *outputBuf, const char *psz, size_t outputSize) const = 0; + virtual size_t WC2MB(char *outputBuf, const wchar_t *psz, size_t outputSize) const = 0; + + // MB <-> WC + const wxWCharBuffer cMB2WC(const char *psz) const; + const wxCharBuffer cWC2MB(const wchar_t *psz) const; + + // MB <-> WC for strings with embedded null characters + // + // pszLen length of the input string + // pOutSize gets the final size of the converted string + const wxWCharBuffer cMB2WC(const char *psz, size_t pszLen, size_t* pOutSize) const; + const wxCharBuffer cWC2MB(const wchar_t *psz, size_t pszLen, size_t* pOutSize) const; + + // convenience functions for converting MB or WC to/from wxWin default #if wxUSE_UNICODE - const wxWCharBuffer cMB2WX(const char *psz) const { return cMB2WC(psz); } - const wxCharBuffer cWX2MB(const wchar_t *psz) const { return cWC2MB(psz); } - const wchar_t* cWC2WX(const wchar_t *psz) const { return psz; } - const wchar_t* cMB2WC(const wchar_t *psz) const { return psz; } -#else - const char* cMB2WX(const char *psz) const { return psz; } - const char* cWX2MB(const char *psz) const { return psz; } - const wxCharBuffer cWC2WX(const wchar_t *psz) const { return cWC2MB(psz); } - const wxWCharBuffer cWX2WC(const char *psz) const { return cMB2WC(psz); } -#endif + const wxWCharBuffer cMB2WX(const char *psz) const { return cMB2WC(psz); } + const wxCharBuffer cWX2MB(const wchar_t *psz) const { return cWC2MB(psz); } + const wchar_t* cWC2WX(const wchar_t *psz) const { return psz; } + const wchar_t* cWX2WC(const wchar_t *psz) const { return psz; } +#else // ANSI + const char* cMB2WX(const char *psz) const { return psz; } + const char* cWX2MB(const char *psz) const { return psz; } + const wxCharBuffer cWC2WX(const wchar_t *psz) const { return cWC2MB(psz); } + const wxWCharBuffer cWX2WC(const char *psz) const { return cMB2WC(psz); } +#endif // Unicode/ANSI + + // virtual dtor for any base class + virtual ~wxMBConv(); }; -WXDLLEXPORT_DATA(extern wxMBConv) wxConvLibc; +// ---------------------------------------------------------------------------- +// wxMBConvLibc uses standard mbstowcs() and wcstombs() functions for +// conversion (hence it depends on the current locale) +// ---------------------------------------------------------------------------- -//--------------------------------------------------------------------------- -// wxMBConvFile (for conversion to filenames) -//--------------------------------------------------------------------------- +class WXDLLIMPEXP_BASE wxMBConvLibc : public wxMBConv +{ +public: + virtual size_t MB2WC(wchar_t *outputBuf, const char *psz, size_t outputSize) const; + virtual size_t WC2MB(char *outputBuf, const wchar_t *psz, size_t outputSize) const; +}; + +#ifdef __UNIX__ -class WXDLLEXPORT wxMBConvFile: public wxMBConv +// ---------------------------------------------------------------------------- +// wxConvBrokenFileNames is made for Unix in Unicode mode when +// files are accidentally written in an encoding which is not +// the system encoding. Typically, the system encoding will be +// UTF8 but there might be files stored in ISO8859-1 on disk. +// ---------------------------------------------------------------------------- + +class WXDLLIMPEXP_BASE wxConvBrokenFileNames : public wxMBConv { public: - virtual size_t MB2WC(wchar_t *buf, const char *psz, size_t n) const; - virtual size_t WC2MB(char *buf, const wchar_t *psz, size_t n) const; + wxConvBrokenFileNames(); + virtual ~wxConvBrokenFileNames() { delete m_conv; } + + virtual size_t MB2WC(wchar_t *outputBuf, const char *psz, size_t outputSize) const; + virtual size_t WC2MB(char *outputBuf, const wchar_t *psz, size_t outputSize) const; + +private: + // the conversion object we forward to + wxMBConv *m_conv; }; -WXDLLEXPORT_DATA(extern wxMBConvFile) wxConvFile; +#endif -//--------------------------------------------------------------------------- +// ---------------------------------------------------------------------------- // wxMBConvUTF7 (for conversion using UTF7 encoding) -//--------------------------------------------------------------------------- +// ---------------------------------------------------------------------------- -class WXDLLEXPORT wxMBConvUTF7: public wxMBConv +class WXDLLIMPEXP_BASE wxMBConvUTF7 : public wxMBConv { public: - virtual size_t MB2WC(wchar_t *buf, const char *psz, size_t n) const; - virtual size_t WC2MB(char *buf, const wchar_t *psz, size_t n) const; + virtual size_t MB2WC(wchar_t *outputBuf, const char *psz, size_t outputSize) const; + virtual size_t WC2MB(char *outputBuf, const wchar_t *psz, size_t outputSize) const; }; -WXDLLEXPORT_DATA(extern wxMBConvUTF7) wxConvUTF7; - -//--------------------------------------------------------------------------- +// ---------------------------------------------------------------------------- // wxMBConvUTF8 (for conversion using UTF8 encoding) -//--------------------------------------------------------------------------- +// ---------------------------------------------------------------------------- -class WXDLLEXPORT wxMBConvUTF8: public wxMBConv +class WXDLLIMPEXP_BASE wxMBConvUTF8 : public wxMBConv { public: - virtual size_t MB2WC(wchar_t *buf, const char *psz, size_t n) const; - virtual size_t WC2MB(char *buf, const wchar_t *psz, size_t n) const; + enum { + MAP_INVALID_UTF8_NOT = 0, + MAP_INVALID_UTF8_TO_PUA = 1, + MAP_INVALID_UTF8_TO_OCTAL = 2 + }; + + wxMBConvUTF8(int options = MAP_INVALID_UTF8_NOT) : m_options(options) { } + virtual size_t MB2WC(wchar_t *outputBuf, const char *psz, size_t outputSize) const; + virtual size_t WC2MB(char *outputBuf, const wchar_t *psz, size_t outputSize) const; + +private: + int m_options; }; -WXDLLEXPORT_DATA(extern wxMBConvUTF8) wxConvUTF8; +// ---------------------------------------------------------------------------- +// wxMBConvUTF16LE (for conversion using UTF16 Little Endian encoding) +// ---------------------------------------------------------------------------- + +class WXDLLIMPEXP_BASE wxMBConvUTF16LE : public wxMBConv +{ +public: + virtual size_t MB2WC(wchar_t *outputBuf, const char *psz, size_t outputSize) const; + virtual size_t WC2MB(char *outputBuf, const wchar_t *psz, size_t outputSize) const; +}; -#ifdef __WXGTK12__ -//--------------------------------------------------------------------------- -// wxMBConvUTF8 (for conversion using GDK's internal converions) -//--------------------------------------------------------------------------- +// ---------------------------------------------------------------------------- +// wxMBConvUTF16BE (for conversion using UTF16 Big Endian encoding) +// ---------------------------------------------------------------------------- -class WXDLLEXPORT wxMBConvGdk: public wxMBConv +class WXDLLIMPEXP_BASE wxMBConvUTF16BE : public wxMBConv { public: - virtual size_t MB2WC(wchar_t *buf, const char *psz, size_t n) const; - virtual size_t WC2MB(char *buf, const wchar_t *psz, size_t n) const; + virtual size_t MB2WC(wchar_t *outputBuf, const char *psz, size_t outputSize) const; + virtual size_t WC2MB(char *outputBuf, const wchar_t *psz, size_t outputSize) const; }; -WXDLLEXPORT_DATA(extern wxMBConvGdk) wxConvGdk; -#endif +// ---------------------------------------------------------------------------- +// wxMBConvUCS4LE (for conversion using UTF32 Little Endian encoding) +// ---------------------------------------------------------------------------- + +class WXDLLIMPEXP_BASE wxMBConvUTF32LE : public wxMBConv +{ +public: + virtual size_t MB2WC(wchar_t *outputBuf, const char *psz, size_t outputSize) const; + virtual size_t WC2MB(char *outputBuf, const wchar_t *psz, size_t outputSize) const; +}; + +// ---------------------------------------------------------------------------- +// wxMBConvUCS4BE (for conversion using UTF32 Big Endian encoding) +// ---------------------------------------------------------------------------- + +class WXDLLIMPEXP_BASE wxMBConvUTF32BE : public wxMBConv +{ +public: + virtual size_t MB2WC(wchar_t *outputBuf, const char *psz, size_t outputSize) const; + virtual size_t WC2MB(char *outputBuf, const wchar_t *psz, size_t outputSize) const; +}; -//--------------------------------------------------------------------------- -// wxCSConv (for conversion based on laodable char sets) -//--------------------------------------------------------------------------- +// ---------------------------------------------------------------------------- +// wxCSConv (for conversion based on loadable char sets) +// ---------------------------------------------------------------------------- -class wxCharacterSet; +#include "wx/fontenc.h" -class WXDLLEXPORT wxCSConv: public wxMBConv +class WXDLLIMPEXP_BASE wxCSConv : public wxMBConv { -private: - wxChar *m_name; - wxCharacterSet *m_cset; - bool m_deferred; - void SetName(const wxChar *charset); - public: - wxCSConv(const wxChar *charset); - virtual ~wxCSConv(); - void LoadNow(); - virtual size_t MB2WC(wchar_t *buf, const char *psz, size_t n) const; - virtual size_t WC2MB(char *buf, const wchar_t *psz, size_t n) const; + // we can be created either from charset name or from an encoding constant + // but we can't have both at once + wxCSConv(const wxChar *charset); + wxCSConv(wxFontEncoding encoding); + + wxCSConv(const wxCSConv& conv); + virtual ~wxCSConv(); + + wxCSConv& operator=(const wxCSConv& conv); + + virtual size_t MB2WC(wchar_t *outputBuf, const char *psz, size_t outputSize) const; + virtual size_t WC2MB(char *outputBuf, const wchar_t *psz, size_t outputSize) const; + + void Clear() ; + +private: + // common part of all ctors + void Init(); + + // creates m_convReal if necessary + void CreateConvIfNeeded() const; + + // do create m_convReal (unconditionally) + wxMBConv *DoCreate() const; + + // set the name (may be only called when m_name == NULL), makes copy of + // charset string + void SetName(const wxChar *charset); + + + // note that we can't use wxString here because of compilation + // dependencies: we're included from wx/string.h + wxChar *m_name; + wxFontEncoding m_encoding; + + // use CreateConvIfNeeded() before accessing m_convReal! + wxMBConv *m_convReal; + bool m_deferred; }; -WXDLLEXPORT_DATA(extern wxCSConv) wxConvLocal; -#define wxConv_local wxConvLocal -WXDLLEXPORT_DATA(extern wxMBConv *) wxConvCurrent; -#define wxConv_current wxConvCurrent +// ---------------------------------------------------------------------------- +// declare predefined conversion objects +// ---------------------------------------------------------------------------- + +// conversion to be used with all standard functions affected by locale, e.g. +// strtol(), strftime(), ... +extern WXDLLIMPEXP_DATA_BASE(wxMBConv&) wxConvLibc; + +// conversion ISO-8859-1/UTF-7/UTF-8 <-> wchar_t +extern WXDLLIMPEXP_DATA_BASE(wxCSConv&) wxConvISO8859_1; +extern WXDLLIMPEXP_DATA_BASE(wxMBConvUTF7&) wxConvUTF7; +extern WXDLLIMPEXP_DATA_BASE(wxMBConvUTF8&) wxConvUTF8; + +// conversion used for the file names on the systems where they're not Unicode +// (basically anything except Windows) +// +// this is used by all file functions, can be changed by the application +// +// by default UTF-8 under Mac OS X and wxConvLibc elsewhere (but it's not used +// under Windows normally) +extern WXDLLIMPEXP_DATA_BASE(wxMBConv *) wxConvFileName; + +// backwards compatible define +#define wxConvFile (*wxConvFileName) + +// the current conversion object, may be set to any conversion, is used by +// default in a couple of places inside wx (initially same as wxConvLibc) +extern WXDLLIMPEXP_DATA_BASE(wxMBConv *) wxConvCurrent; -//--------------------------------------------------------------------------- +// ??? +extern WXDLLIMPEXP_DATA_BASE(wxCSConv&) wxConvLocal; + + +// ---------------------------------------------------------------------------- +// endianness-dependent conversions +// ---------------------------------------------------------------------------- + +#ifdef WORDS_BIGENDIAN + typedef wxMBConvUTF16BE wxMBConvUTF16; + typedef wxMBConvUTF32BE wxMBConvUTF32; +#else + typedef wxMBConvUTF16LE wxMBConvUTF16; + typedef wxMBConvUTF32LE wxMBConvUTF32; +#endif + +// ---------------------------------------------------------------------------- // filename conversion macros -//--------------------------------------------------------------------------- +// ---------------------------------------------------------------------------- // filenames are multibyte on Unix and probably widechar on Windows? -#if defined(__UNIX__) || defined(__BORLANDC__) -#define wxMBFILES 1 +#if defined(__UNIX__) || defined(__BORLANDC__) || defined(__WXMAC__ ) + #define wxMBFILES 1 #else -#define wxMBFILES 0 + #define wxMBFILES 0 #endif -#if wxMBFILES -#define wxFNCONV(name) wxConvFile.cWX2MB(name) -#define FNSTRINGCAST MBSTRINGCAST +#if wxMBFILES && wxUSE_UNICODE + #define wxFNCONV(name) wxConvFileName->cWX2MB(name) + #define wxFNSTRINGCAST wxMBSTRINGCAST #else -#define wxFNCONV(name) name -#define FNSTRINGCAST WXSTRINGCAST +#if defined( __WXOSX__ ) && wxMBFILES + #define wxFNCONV(name) wxConvFileName->cWC2MB( wxConvLocal.cWX2WC(name) ) +#else + #define wxFNCONV(name) name +#endif + #define wxFNSTRINGCAST WXSTRINGCAST #endif -#else - // !wxUSE_WCHAR_T +#else // !wxUSE_WCHAR_T -//--------------------------------------------------------------------------- +// ---------------------------------------------------------------------------- // stand-ins in absence of wchar_t -//--------------------------------------------------------------------------- +// ---------------------------------------------------------------------------- -class WXDLLEXPORT wxMBConv +class WXDLLIMPEXP_BASE wxMBConv { public: - const char* cMB2WX(const char *psz) const { return psz; } - const char* cWX2MB(const char *psz) const { return psz; } + const char* cMB2WX(const char *psz) const { return psz; } + const char* cWX2MB(const char *psz) const { return psz; } }; -WXDLLEXPORT_DATA(extern wxMBConv) wxConvLibc, wxConvFile; -WXDLLEXPORT_DATA(extern wxMBConv *) wxConvCurrent; + +#define wxConvFile wxConvLocal + +extern WXDLLIMPEXP_DATA_BASE(wxMBConv) wxConvLibc, + wxConvLocal, + wxConvISO8859_1, + wxConvUTF8; +extern WXDLLIMPEXP_DATA_BASE(wxMBConv *) wxConvCurrent; #define wxFNCONV(name) name -#define FNSTRINGCAST WXSTRINGCAST +#define wxFNSTRINGCAST WXSTRINGCAST #endif // wxUSE_WCHAR_T -#endif +// ---------------------------------------------------------------------------- +// macros for the most common conversions +// ---------------------------------------------------------------------------- + +#if wxUSE_UNICODE + #define wxConvertWX2MB(s) wxConvCurrent->cWX2MB(s) + #define wxConvertMB2WX(s) wxConvCurrent->cMB2WX(s) +#else // ANSI + // no conversions to do + #define wxConvertWX2MB(s) (s) + #define wxConvertMB2WX(s) (s) +#endif // Unicode/ANSI + +#endif // _WX_WXSTRCONVH__