X-Git-Url: https://git.saurik.com/wxWidgets.git/blobdiff_plain/d51305e81eb593bedd92cda864671e15209c268f..4e0a055757f928dce8c53b8e4cb60a699ab43e07:/include/wx/strconv.h?ds=sidebyside diff --git a/include/wx/strconv.h b/include/wx/strconv.h index ce0f040121..b36ad0b862 100644 --- a/include/wx/strconv.h +++ b/include/wx/strconv.h @@ -1,23 +1,20 @@ /////////////////////////////////////////////////////////////////////////////// // Name: strconv.h // Purpose: conversion routines for char sets any Unicode -// Author: Robert Roebling, Ove Kaaven +// Author: Ove Kaaven, Robert Roebling, Vadim Zeitlin // Modified by: // Created: 29/01/98 // RCS-ID: $Id$ -// Copyright: (c) 1998 Ove Kaaven, Robert Roebling, Vadim Zeitlin +// Copyright: (c) 1998 Ove Kaaven, Robert Roebling +// (c) 1998-2006 Vadim Zeitlin // Licence: wxWindows licence /////////////////////////////////////////////////////////////////////////////// -#ifndef _WX_WXSTRCONVH__ -#define _WX_WXSTRCONVH__ - -#if defined(__GNUG__) && !defined(NO_GCC_PRAGMA) - #pragma interface "strconv.h" -#endif +#ifndef _WX_STRCONV_H_ +#define _WX_STRCONV_H_ #include "wx/defs.h" -#include "wx/wxchar.h" +#include "wx/chartype.h" #include "wx/buffer.h" #ifdef __DIGITALMARS__ @@ -32,30 +29,83 @@ #if wxUSE_WCHAR_T +class WXDLLIMPEXP_BASE wxString; + +// the error value returned by wxMBConv methods +#define wxCONV_FAILED ((size_t)-1) + +// the default value for some length parameters meaning that the string is +// NUL-terminated +#define wxNO_LEN ((size_t)-1) + // ---------------------------------------------------------------------------- // wxMBConv (abstract base class for conversions) // ---------------------------------------------------------------------------- +// When deriving a new class from wxMBConv you must reimplement ToWChar() and +// FromWChar() methods which are not pure virtual only for historical reasons, +// don't let the fact that the existing classes implement MB2WC/WC2MB() instead +// confuse you. +// +// You also have to implement Clone() to allow copying the conversions +// polymorphically. +// +// And you might need to override GetMBNulLen() as well. class WXDLLIMPEXP_BASE wxMBConv { public: - // the actual conversion takes place here + // The functions doing actual conversion from/to narrow to/from wide + // character strings. + // + // On success, the return value is the length (i.e. the number of + // characters, not bytes) of the converted string including any trailing + // L'\0' or (possibly multiple) '\0'(s). If the conversion fails or if + // there is not enough space for everything, including the trailing NUL + // character(s), in the output buffer, wxCONV_FAILED is returned. // - // 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; + // In the special case when dstLen is 0 (outputBuf may be NULL then) the + // return value is the length of the needed buffer but nothing happens + // otherwise. If srcLen is wxNO_LEN, the entire string, up to and + // including the trailing NUL(s), is converted, otherwise exactly srcLen + // bytes are. + // + // Typical usage: + // + // size_t dstLen = conv.ToWChar(NULL, 0, src); + // if ( dstLen != wxCONV_FAILED ) + // ... handle error ... + // wchar_t *wbuf = new wchar_t[dstLen]; + // conv.ToWChar(wbuf, dstLen, src); + // + virtual size_t ToWChar(wchar_t *dst, size_t dstLen, + const char *src, size_t srcLen = wxNO_LEN) const; + + virtual size_t FromWChar(char *dst, size_t dstLen, + const wchar_t *src, size_t srcLen = wxNO_LEN) const; - // 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 + // Convenience functions for translating NUL-terminated strings: returns + // the buffer containing the converted string or NULL pointer if the + // conversion failed. + const wxWCharBuffer cMB2WC(const char *in) const; + const wxCharBuffer cWC2MB(const wchar_t *in) const; + + // Convenience functions for converting strings which may contain embedded + // NULs and don't have to be NUL-terminated. + // + // inLen is the length of the buffer including trailing NUL if any: if the + // last 4 bytes of the buffer are all NULs, these functions are more + // efficient as they avoid copying the string, but otherwise a copy is made + // internally which could be quite bad for (very) long strings. // - // 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; + // outLen receives, if not NULL, the length of the converted string or 0 if + // the conversion failed (returning 0 and not -1 in this case makes it + // difficult to distinguish between failed conversion and empty input but + // this is done for backwards compatibility) + const wxWCharBuffer + cMB2WC(const char *in, size_t inLen, size_t *outLen) const; + const wxCharBuffer + cWC2MB(const wchar_t *in, size_t inLen, size_t *outLen) const; // convenience functions for converting MB or WC to/from wxWin default #if wxUSE_UNICODE @@ -70,6 +120,53 @@ public: const wxWCharBuffer cWX2WC(const char *psz) const { return cMB2WC(psz); } #endif // Unicode/ANSI + // this function is used in the implementation of cMB2WC() to distinguish + // between the following cases: + // + // a) var width encoding with strings terminated by a single NUL + // (usual multibyte encodings): return 1 in this case + // b) fixed width encoding with 2 bytes/char and so terminated by + // 2 NULs (UTF-16/UCS-2 and variants): return 2 in this case + // c) fixed width encoding with 4 bytes/char and so terminated by + // 4 NULs (UTF-32/UCS-4 and variants): return 4 in this case + // + // anything else is not supported currently and -1 should be returned + virtual size_t GetMBNulLen() const { return 1; } + + // return the maximal value currently returned by GetMBNulLen() for any + // encoding + static size_t GetMaxMBNulLen() { return 4 /* for UTF-32 */; } + +#if wxUSE_UNICODE_UTF8 + // return true if the converter's charset is UTF-8, i.e. char* strings + // decoded using this object can be directly copied to wxString's internal + // storage without converting to WC and than back to UTF-8 MB string + virtual bool IsUTF8() const { return false; } +#endif + + // The old conversion functions. The existing classes currently mostly + // implement these ones but we're in transition to using To/FromWChar() + // instead and any new classes should implement just the new functions. + // For now, however, we provide default implementation of To/FromWChar() in + // this base class in terms of MB2WC/WC2MB() to avoid having to rewrite all + // the conversions at once. + // + // On success, the return value is the length (i.e. the number of + // characters, not bytes) not counting the trailing NUL(s) of the converted + // string. On failure, (size_t)-1 is returned. In the special case when + // outputBuf is NULL the return value is the same one but nothing is + // written to the buffer. + // + // Note that outLen is the length of the output buffer, not the length of + // the input (which is always supposed to be terminated by one or more + // NULs, as appropriate for the encoding)! + virtual size_t MB2WC(wchar_t *out, const char *in, size_t outLen) const; + virtual size_t WC2MB(char *out, const wchar_t *in, size_t outLen) const; + + + // make a heap-allocated copy of this object + virtual wxMBConv *Clone() const = 0; + // virtual dtor for any base class virtual ~wxMBConv(); }; @@ -84,8 +181,65 @@ 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; + + virtual wxMBConv *Clone() const { return new wxMBConvLibc; } + +#if wxUSE_UNICODE_UTF8 + virtual bool IsUTF8() const { return wxLocaleIsUtf8; } +#endif }; +#ifdef __UNIX__ + +// ---------------------------------------------------------------------------- +// 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: + wxConvBrokenFileNames(const wxString& charset); + wxConvBrokenFileNames(const wxConvBrokenFileNames& conv) + : wxMBConv(), + m_conv(conv.m_conv ? conv.m_conv->Clone() : NULL) + { + } + virtual ~wxConvBrokenFileNames() { delete m_conv; } + + virtual size_t MB2WC(wchar_t *out, const char *in, size_t outLen) const + { + return m_conv->MB2WC(out, in, outLen); + } + + virtual size_t WC2MB(char *out, const wchar_t *in, size_t outLen) const + { + return m_conv->WC2MB(out, in, outLen); + } + + virtual size_t GetMBNulLen() const + { + // cast needed to call a private function + return m_conv->GetMBNulLen(); + } + +#if wxUSE_UNICODE_UTF8 + virtual bool IsUTF8() const { return m_conv->IsUTF8(); } +#endif + + virtual wxMBConv *Clone() const { return new wxConvBrokenFileNames(*this); } + +private: + // the conversion object we forward to + wxMBConv *m_conv; + + DECLARE_NO_ASSIGN_CLASS(wxConvBrokenFileNames) +}; + +#endif // __UNIX__ + // ---------------------------------------------------------------------------- // wxMBConvUTF7 (for conversion using UTF7 encoding) // ---------------------------------------------------------------------------- @@ -95,6 +249,8 @@ class WXDLLIMPEXP_BASE wxMBConvUTF7 : 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; + + virtual wxMBConv *Clone() const { return new wxMBConvUTF7; } }; // ---------------------------------------------------------------------------- @@ -104,7 +260,10 @@ public: class WXDLLIMPEXP_BASE wxMBConvUTF8 : public wxMBConv { public: - enum { + // FIXME-UTF8: split this class into multiple classes, one strict and + // other lossy (PUA, OCTAL mappings) + enum + { MAP_INVALID_UTF8_NOT = 0, MAP_INVALID_UTF8_TO_PUA = 1, MAP_INVALID_UTF8_TO_OCTAL = 2 @@ -113,53 +272,110 @@ public: 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; - + + virtual wxMBConv *Clone() const { return new wxMBConvUTF8(m_options); } + +#if wxUSE_UNICODE_UTF8 + // NB: other mapping modes are not, strictly speaking, UTF-8, so we can't + // take the shortcut in that case + virtual bool IsUTF8() const { return m_options == MAP_INVALID_UTF8_NOT; } +#endif + private: int m_options; }; +// ---------------------------------------------------------------------------- +// wxMBConvUTF16Base: for both LE and BE variants +// ---------------------------------------------------------------------------- + +class WXDLLIMPEXP_BASE wxMBConvUTF16Base : public wxMBConv +{ +public: + enum { BYTES_PER_CHAR = 2 }; + + virtual size_t GetMBNulLen() const { return BYTES_PER_CHAR; } + +protected: + // return the length of the buffer using srcLen if it's not wxNO_LEN and + // computing the length ourselves if it is; also checks that the length is + // even if specified as we need an entire number of UTF-16 characters and + // returns wxNO_LEN which indicates error if it is odd + static size_t GetLength(const char *src, size_t srcLen); +}; + // ---------------------------------------------------------------------------- // wxMBConvUTF16LE (for conversion using UTF16 Little Endian encoding) // ---------------------------------------------------------------------------- -class WXDLLIMPEXP_BASE wxMBConvUTF16LE : public wxMBConv +class WXDLLIMPEXP_BASE wxMBConvUTF16LE : public wxMBConvUTF16Base { 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; + virtual size_t ToWChar(wchar_t *dst, size_t dstLen, + const char *src, size_t srcLen = wxNO_LEN) const; + virtual size_t FromWChar(char *dst, size_t dstLen, + const wchar_t *src, size_t srcLen = wxNO_LEN) const; + virtual wxMBConv *Clone() const { return new wxMBConvUTF16LE; } }; // ---------------------------------------------------------------------------- // wxMBConvUTF16BE (for conversion using UTF16 Big Endian encoding) // ---------------------------------------------------------------------------- -class WXDLLIMPEXP_BASE wxMBConvUTF16BE : public wxMBConv +class WXDLLIMPEXP_BASE wxMBConvUTF16BE : public wxMBConvUTF16Base { 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; + virtual size_t ToWChar(wchar_t *dst, size_t dstLen, + const char *src, size_t srcLen = wxNO_LEN) const; + virtual size_t FromWChar(char *dst, size_t dstLen, + const wchar_t *src, size_t srcLen = wxNO_LEN) const; + virtual wxMBConv *Clone() const { return new wxMBConvUTF16BE; } }; // ---------------------------------------------------------------------------- -// wxMBConvUCS4LE (for conversion using UTF32 Little Endian encoding) +// wxMBConvUTF32Base: base class for both LE and BE variants // ---------------------------------------------------------------------------- -class WXDLLIMPEXP_BASE wxMBConvUTF32LE : public wxMBConv +class WXDLLIMPEXP_BASE wxMBConvUTF32Base : 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; + enum { BYTES_PER_CHAR = 4 }; + + virtual size_t GetMBNulLen() const { return BYTES_PER_CHAR; } + +protected: + // this is similar to wxMBConvUTF16Base method with the same name except + // that, of course, it verifies that length is divisible by 4 if given and + // not by 2 + static size_t GetLength(const char *src, size_t srcLen); }; // ---------------------------------------------------------------------------- -// wxMBConvUCS4BE (for conversion using UTF32 Big Endian encoding) +// wxMBConvUTF32LE (for conversion using UTF32 Little Endian encoding) // ---------------------------------------------------------------------------- -class WXDLLIMPEXP_BASE wxMBConvUTF32BE : public wxMBConv +class WXDLLIMPEXP_BASE wxMBConvUTF32LE : public wxMBConvUTF32Base { 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; + virtual size_t ToWChar(wchar_t *dst, size_t dstLen, + const char *src, size_t srcLen = wxNO_LEN) const; + virtual size_t FromWChar(char *dst, size_t dstLen, + const wchar_t *src, size_t srcLen = wxNO_LEN) const; + virtual wxMBConv *Clone() const { return new wxMBConvUTF32LE; } +}; + +// ---------------------------------------------------------------------------- +// wxMBConvUTF32BE (for conversion using UTF32 Big Endian encoding) +// ---------------------------------------------------------------------------- + +class WXDLLIMPEXP_BASE wxMBConvUTF32BE : public wxMBConvUTF32Base +{ +public: + virtual size_t ToWChar(wchar_t *dst, size_t dstLen, + const char *src, size_t srcLen = wxNO_LEN) const; + virtual size_t FromWChar(char *dst, size_t dstLen, + const wchar_t *src, size_t srcLen = wxNO_LEN) const; + virtual wxMBConv *Clone() const { return new wxMBConvUTF32BE; } }; // ---------------------------------------------------------------------------- @@ -173,7 +389,7 @@ class WXDLLIMPEXP_BASE wxCSConv : public wxMBConv public: // 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(const wxString& charset); wxCSConv(wxFontEncoding encoding); wxCSConv(const wxCSConv& conv); @@ -181,10 +397,24 @@ public: wxCSConv& operator=(const wxCSConv& conv); + virtual size_t ToWChar(wchar_t *dst, size_t dstLen, + const char *src, size_t srcLen = wxNO_LEN) const; + virtual size_t FromWChar(char *dst, size_t dstLen, + const wchar_t *src, size_t srcLen = wxNO_LEN) 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; + virtual size_t GetMBNulLen() const; + +#if wxUSE_UNICODE_UTF8 + virtual bool IsUTF8() const; +#endif + + virtual wxMBConv *Clone() const { return new wxCSConv(*this); } - void Clear() ; + void Clear(); + + // return true if the conversion could be initilized successfully + bool IsOk() const; private: // common part of all ctors @@ -198,12 +428,12 @@ private: // set the name (may be only called when m_name == NULL), makes copy of // charset string - void SetName(const wxChar *charset); + void SetName(const char *charset); // note that we can't use wxString here because of compilation // dependencies: we're included from wx/string.h - wxChar *m_name; + char *m_name; wxFontEncoding m_encoding; // use CreateConvIfNeeded() before accessing m_convReal! @@ -211,43 +441,40 @@ private: bool m_deferred; }; -#ifdef __UNIX__ // ---------------------------------------------------------------------------- -// wxConvBrokenFileNames is made for GTK2 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. +// declare predefined conversion objects // ---------------------------------------------------------------------------- -class wxConvBrokenFileNames : public wxMBConv -{ -public: - wxConvBrokenFileNames(); - virtual ~wxConvBrokenFileNames() { delete m_conv; } +// Note: this macro is an implementation detail (see the comment in +// strconv.cpp). The wxGet_XXX() and wxGet_XXXPtr() functions shouldn't be +// used by user code and neither should XXXPtr, use the wxConvXXX macro +// instead. +#define WX_DECLARE_GLOBAL_CONV(klass, name) \ + extern WXDLLIMPEXP_DATA_BASE(klass*) name##Ptr; \ + extern WXDLLIMPEXP_BASE klass* wxGet_##name##Ptr(); \ + inline klass& wxGet_##name() \ + { \ + if ( !name##Ptr ) \ + name##Ptr = wxGet_##name##Ptr(); \ + return *name##Ptr; \ + } - 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; -}; -#endif -// __UNIX__ - -// ---------------------------------------------------------------------------- -// 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; +WX_DECLARE_GLOBAL_CONV(wxMBConv, wxConvLibc) +#define wxConvLibc wxGet_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; +WX_DECLARE_GLOBAL_CONV(wxCSConv, wxConvISO8859_1) +#define wxConvISO8859_1 wxGet_wxConvISO8859_1() + +WX_DECLARE_GLOBAL_CONV(wxMBConvUTF8, wxConvUTF8) +#define wxConvUTF8 wxGet_wxConvUTF8() + +WX_DECLARE_GLOBAL_CONV(wxMBConvUTF7, wxConvUTF7) +#define wxConvUTF7 wxGet_wxConvUTF7() // conversion used for the file names on the systems where they're not Unicode // (basically anything except Windows) @@ -265,9 +492,17 @@ extern WXDLLIMPEXP_DATA_BASE(wxMBConv *) wxConvFileName; // default in a couple of places inside wx (initially same as wxConvLibc) extern WXDLLIMPEXP_DATA_BASE(wxMBConv *) wxConvCurrent; -// ??? -extern WXDLLIMPEXP_DATA_BASE(wxCSConv&) wxConvLocal; +// the conversion corresponding to the current locale +WX_DECLARE_GLOBAL_CONV(wxCSConv, wxConvLocal) +#define wxConvLocal wxGet_wxConvLocal() + +// the conversion corresponding to the encoding of the standard UI elements +// +// by default this is the same as wxConvLocal but may be changed if the program +// needs to use a fixed encoding +extern WXDLLIMPEXP_DATA_BASE(wxMBConv *) wxConvUI; +#undef WX_DECLARE_GLOBAL_CONV // ---------------------------------------------------------------------------- // endianness-dependent conversions @@ -285,13 +520,7 @@ extern WXDLLIMPEXP_DATA_BASE(wxCSConv&) wxConvLocal; // filename conversion macros // ---------------------------------------------------------------------------- -// filenames are multibyte on Unix and probably widechar on Windows? -#if defined(__UNIX__) || defined(__BORLANDC__) || defined(__WXMAC__ ) - #define wxMBFILES 1 -#else - #define wxMBFILES 0 -#endif - +// filenames are multibyte on Unix and widechar on Windows #if wxMBFILES && wxUSE_UNICODE #define wxFNCONV(name) wxConvFileName->cWX2MB(name) #define wxFNSTRINGCAST wxMBSTRINGCAST @@ -338,12 +567,26 @@ extern WXDLLIMPEXP_DATA_BASE(wxMBConv *) wxConvCurrent; #if wxUSE_UNICODE #define wxConvertWX2MB(s) wxConvCurrent->cWX2MB(s) #define wxConvertMB2WX(s) wxConvCurrent->cMB2WX(s) + + // these functions should be used when the conversions really, really have + // to succeed (usually because we pass their results to a standard C + // function which would crash if we passed NULL to it), so these functions + // always return a valid pointer if their argument is non-NULL + + // this function safety is achieved by trying wxConvLibc first, wxConvUTF8 + // next if it fails and, finally, wxConvISO8859_1 which always succeeds + extern WXDLLIMPEXP_BASE wxWCharBuffer wxSafeConvertMB2WX(const char *s); + + // this function uses wxConvLibc and wxConvUTF8(MAP_INVALID_UTF8_TO_OCTAL) + // if it fails + extern WXDLLIMPEXP_BASE wxCharBuffer wxSafeConvertWX2MB(const wchar_t *ws); #else // ANSI // no conversions to do #define wxConvertWX2MB(s) (s) #define wxConvertMB2WX(s) (s) + #define wxSafeConvertMB2WX(s) (s) + #define wxSafeConvertWX2MB(s) (s) #endif // Unicode/ANSI -#endif - // _WX_WXSTRCONVH__ +#endif // _WX_STRCONV_H_