X-Git-Url: https://git.saurik.com/wxWidgets.git/blobdiff_plain/4f61e22c77b1a7cb148f6567996d816f72ac854f..7873ca319146f39b4685e8727fde91cf601f4bbe:/src/common/strconv.cpp diff --git a/src/common/strconv.cpp b/src/common/strconv.cpp index 17a15ca5a8..ff16d13d00 100644 --- a/src/common/strconv.cpp +++ b/src/common/strconv.cpp @@ -5,8 +5,9 @@ // Modified by: // Created: 29/01/98 // RCS-ID: $Id$ -// Copyright: (c) 1999 Ove Kaaven, Robert Roebling, Vadim Zeitlin, Vaclav Slavik -// Licence: wxWindows license +// Copyright: (c) 1999 Ove Kaaven, Robert Roebling, Vaclav Slavik +// (c) 2000-2003 Vadim Zeitlin +// Licence: wxWindows licence ///////////////////////////////////////////////////////////////////////////// // ============================================================================ @@ -17,7 +18,7 @@ // headers // ---------------------------------------------------------------------------- -#ifdef __GNUG__ +#if defined(__GNUG__) && !defined(NO_GCC_PRAGMA) #pragma implementation "strconv.h" #endif @@ -28,55 +29,65 @@ #pragma hdrstop #endif +#ifndef WX_PRECOMP + #include "wx/intl.h" + #include "wx/log.h" +#endif // WX_PRECOMP + +#include "wx/strconv.h" + +#if wxUSE_WCHAR_T + #ifdef __WXMSW__ - #include "wx/msw/private.h" + #include "wx/msw/private.h" #endif +#ifdef __WINDOWS__ + #include "wx/msw/missing.h" +#endif + +#ifndef __WXWINCE__ #include +#endif + #include #include #include - -#include "wx/debug.h" -#include "wx/strconv.h" -#include "wx/intl.h" -#include "wx/log.h" +#if defined(__WIN32__) && !defined(__WXMICROWIN__) + #define wxHAVE_WIN32_MB2WC +#endif // __WIN32__ but !__WXMICROWIN__ // ---------------------------------------------------------------------------- -// globals +// headers // ---------------------------------------------------------------------------- -WXDLLEXPORT_DATA(wxMBConv *) wxConvCurrent = &wxConvLibc; - - -// ============================================================================ -// implementation -// ============================================================================ - -#if wxUSE_WCHAR_T - #ifdef __SALFORDC__ - #include + #include #endif #ifdef HAVE_ICONV - #include + #include #endif -#ifdef __WXMSW__ - #include +#include "wx/encconv.h" +#include "wx/fontmap.h" +#include "wx/utils.h" + +#ifdef __WXMAC__ +#include +#include +#include + +#include "wx/mac/private.h" // includes mac headers #endif +// ---------------------------------------------------------------------------- +// macros +// ---------------------------------------------------------------------------- #define BSWAP_UCS4(str, len) { unsigned _c; for (_c=0; _c=0x110000) @@ -117,14 +135,14 @@ static size_t encode_utf16(wxUint32 input, wchar_t *output) { if (output) { - *output++ = (input >> 10)+0xd7c0; - *output++ = (input&0x3ff)+0xdc00; + *output++ = (wxUint16) ((input >> 10)+0xd7c0); + *output = (wxUint16) ((input&0x3ff)+0xdc00); } return 2; } } -static size_t decode_utf16(const wchar_t* input, wxUint32& output) +static size_t decode_utf16(const wxUint16* input, wxUint32& output) { if ((*input<0xd800) || (*input>0xdfff)) { @@ -143,119 +161,78 @@ static size_t decode_utf16(const wchar_t* input, wxUint32& output) } } -#endif // WC_UTF16 // ---------------------------------------------------------------------------- // wxMBConv // ---------------------------------------------------------------------------- -WXDLLEXPORT_DATA(wxMBConv) wxConvLibc; - -size_t wxMBConv::MB2WC(wchar_t *buf, const char *psz, size_t n) const -{ - return wxMB2WC(buf, psz, n); -} - -size_t wxMBConv::WC2MB(char *buf, const wchar_t *psz, size_t n) const +wxMBConv::~wxMBConv() { - return wxWC2MB(buf, psz, n); + // nothing to do here } const wxWCharBuffer wxMBConv::cMB2WC(const char *psz) const { - if (psz) + if ( psz ) { - size_t nLen = MB2WC((wchar_t *) NULL, psz, 0); - if (nLen == (size_t)-1) - return wxWCharBuffer((wchar_t *) NULL); - wxWCharBuffer buf(nLen); - MB2WC((wchar_t *)(const wchar_t *) buf, psz, nLen); - return buf; + // calculate the length of the buffer needed first + size_t nLen = MB2WC(NULL, psz, 0); + if ( nLen != (size_t)-1 ) + { + // now do the actual conversion + wxWCharBuffer buf(nLen); + nLen = MB2WC(buf.data(), psz, nLen + 1); // with the trailing NULL + if ( nLen != (size_t)-1 ) + { + return buf; + } + } } - else - return wxWCharBuffer((wchar_t *) NULL); + + wxWCharBuffer buf((wchar_t *)NULL); + + return buf; } -const wxCharBuffer wxMBConv::cWC2MB(const wchar_t *psz) const +const wxCharBuffer wxMBConv::cWC2MB(const wchar_t *pwz) const { - if (psz) + if ( pwz ) { - size_t nLen = WC2MB((char *) NULL, psz, 0); - if (nLen == (size_t)-1) - return wxCharBuffer((char *) NULL); - wxCharBuffer buf(nLen); - WC2MB((char *)(const char *) buf, psz, nLen); - return buf; + size_t nLen = WC2MB(NULL, pwz, 0); + if ( nLen != (size_t)-1 ) + { + wxCharBuffer buf(nLen+3); // space for a wxUint32 trailing zero + nLen = WC2MB(buf.data(), pwz, nLen + 4); + if ( nLen != (size_t)-1 ) + { + return buf; + } + } } - else - return wxCharBuffer((char *) NULL); -} -// ---------------------------------------------------------------------------- -// standard file conversion -// ---------------------------------------------------------------------------- + wxCharBuffer buf((char *)NULL); -WXDLLEXPORT_DATA(wxMBConvFile) wxConvFile; - -// just use the libc conversion for now -size_t wxMBConvFile::MB2WC(wchar_t *buf, const char *psz, size_t n) const -{ - return wxMB2WC(buf, psz, n); -} - -size_t wxMBConvFile::WC2MB(char *buf, const wchar_t *psz, size_t n) const -{ - return wxWC2MB(buf, psz, n); + return buf; } // ---------------------------------------------------------------------------- -// standard gdk conversion +// wxMBConvLibc // ---------------------------------------------------------------------------- -#ifdef __WXGTK12__ - -WXDLLEXPORT_DATA(wxMBConvGdk) wxConvGdk; - -#include - -size_t wxMBConvGdk::MB2WC(wchar_t *buf, const char *psz, size_t n) const +size_t wxMBConvLibc::MB2WC(wchar_t *buf, const char *psz, size_t n) const { - if (buf) - { - return gdk_mbstowcs((GdkWChar *)buf, psz, n); - } - else - { - GdkWChar *nbuf = new GdkWChar[n=strlen(psz)]; - size_t len = gdk_mbstowcs(nbuf, psz, n); - delete[] nbuf; - return len; - } + return wxMB2WC(buf, psz, n); } -size_t wxMBConvGdk::WC2MB(char *buf, const wchar_t *psz, size_t n) const +size_t wxMBConvLibc::WC2MB(char *buf, const wchar_t *psz, size_t n) const { - char *mbstr = gdk_wcstombs((GdkWChar *)psz); - size_t len = mbstr ? strlen(mbstr) : 0; - if (buf) - { - if (len > n) - len = n; - memcpy(buf, psz, len); - if (len < n) - buf[len] = 0; - } - return len; + return wxWC2MB(buf, psz, n); } -#endif // GTK > 1.0 - // ---------------------------------------------------------------------------- // UTF-7 // ---------------------------------------------------------------------------- -WXDLLEXPORT_DATA(wxMBConvUTF7) wxConvUTF7; - #if 0 static char utf7_setD[]="ABCDEFGHIJKLMNOPQRSTUVWXYZ" "abcdefghijklmnopqrstuvwxyz" @@ -285,8 +262,6 @@ size_t wxMBConvUTF7::WC2MB(char * WXUNUSED(buf), // UTF-8 // ---------------------------------------------------------------------------- -WXDLLEXPORT_DATA(wxMBConvUTF8) wxConvUTF8; - static wxUint32 utf8_max[]= { 0x7f, 0x7ff, 0xffff, 0x1fffff, 0x3ffffff, 0x7fffffff, 0xffffffff }; @@ -335,17 +310,18 @@ size_t wxMBConvUTF8::MB2WC(wchar_t *buf, const char *psz, size_t n) const return (size_t)-1; } #ifdef WC_UTF16 - size_t pa = encode_utf16(res, buf); + // cast is ok because wchar_t == wxUuint16 if WC_UTF16 + size_t pa = encode_utf16(res, (wxUint16 *)buf); if (pa == (size_t)-1) return (size_t)-1; if (buf) buf += pa; len += pa; -#else +#else // !WC_UTF16 if (buf) *buf++ = res; len++; -#endif +#endif // WC_UTF16/!WC_UTF16 } } } @@ -362,7 +338,8 @@ size_t wxMBConvUTF8::WC2MB(char *buf, const wchar_t *psz, size_t n) const { wxUint32 cc; #ifdef WC_UTF16 - size_t pa = decode_utf16(psz, cc); + // cast is ok for WC_UTF16 + size_t pa = decode_utf16((const wxUint16 *)psz, cc); psz += (pa == (size_t)-1) ? 1 : pa; #else cc=(*psz++) & 0x7fffffff; @@ -373,7 +350,7 @@ size_t wxMBConvUTF8::WC2MB(char *buf, const wchar_t *psz, size_t n) const { // plain ASCII char if (buf) - *buf++ = cc; + *buf++ = (char) cc; len++; } @@ -382,162 +359,484 @@ size_t wxMBConvUTF8::WC2MB(char *buf, const wchar_t *psz, size_t n) const len += cnt + 1; if (buf) { - *buf++ = (-128 >> cnt) | ((cc >> (cnt * 6)) & (0x3f >> cnt)); + *buf++ = (char) ((-128 >> cnt) | ((cc >> (cnt * 6)) & (0x3f >> cnt))); while (cnt--) - *buf++ = 0x80 | ((cc >> (cnt * 6)) & 0x3f); + *buf++ = (char) (0x80 | ((cc >> (cnt * 6)) & 0x3f)); } } } if (buf && (len wxConvUTF8) -// - move wxEncodingConverter meat in here +#ifdef WC_UTF16 -#if defined(__WIN32__) && !defined(__WXMICROWIN__) +// copy 16bit MB to 16bit String +size_t wxMBConvUTF16straight::MB2WC(wchar_t *buf, const char *psz, size_t n) const +{ + size_t len=0; + + while (*(wxUint16*)psz && (!buf || len < n)) + { + if (buf) + *buf++ = *(wxUint16*)psz; + len++; + + psz += sizeof(wxUint16); + } + if (buf && lenCharsetToEncoding(name, FALSE); - if ( enc == wxFONTENCODING_SYSTEM ) - return -1; + while (*psz && (!buf || len < n)) + { + if (buf) + { + *buf++ = ((char*)psz)[1]; + *buf++ = ((char*)psz)[0]; + } + len += sizeof(wxUint16); + psz++; + } + if (buf && len<=n-sizeof(wxUint16)) *(wxUint16*)buf=0; - // the use the helper function - return wxEncodingToCodepage(enc); + return len; } -#endif // wxUSE_FONTMAP -#endif // wxUSE_GUI +#else // WC_UTF16 + + +// copy 16bit MB to 32bit String +size_t wxMBConvUTF16straight::MB2WC(wchar_t *buf, const char *psz, size_t n) const +{ + size_t len=0; + + while (*(wxUint16*)psz && (!buf || len < n)) + { + wxUint32 cc; + size_t pa=decode_utf16((wxUint16*)psz, cc); + if (pa == (size_t)-1) + return pa; + + if (buf) + *buf++ = cc; + len++; + psz += pa * sizeof(wxUint16); + } + if (buf && len 1) + { + *(wxUint16*)buf = cc[1]; + buf += sizeof(wxUint16); + } + } + + len += pa*sizeof(wxUint16); + psz++; + } + if (buf && len<=n-sizeof(wxUint16)) *(wxUint16*)buf=0; + + return len; +} + + +// swap 16bit MB to 32bit String +size_t wxMBConvUTF16swap::MB2WC(wchar_t *buf, const char *psz, size_t n) const +{ + size_t len=0; + + while (*(wxUint16*)psz && (!buf || len < n)) + { + wxUint32 cc; + char tmp[4]; + tmp[0]=psz[1]; tmp[1]=psz[0]; + tmp[2]=psz[3]; tmp[3]=psz[2]; + + size_t pa=decode_utf16((wxUint16*)tmp, cc); + if (pa == (size_t)-1) + return pa; + + if (buf) + *buf++ = cc; + + len++; + psz += pa * sizeof(wxUint16); + } + if (buf && len 1) + { + *buf++ = ((char*)cc)[3]; + *buf++ = ((char*)cc)[2]; + } + } + + len += pa*sizeof(wxUint16); + psz++; + } + if (buf && len<=n-sizeof(wxUint16)) *(wxUint16*)buf=0; + + return len; +} + +#endif // WC_UTF16 + + +// ---------------------------------------------------------------------------- +// UTF-32 +// ---------------------------------------------------------------------------- + +#ifdef WORDS_BIGENDIAN +#define wxMBConvUTF32straight wxMBConvUTF32BE +#define wxMBConvUTF32swap wxMBConvUTF32LE +#else +#define wxMBConvUTF32swap wxMBConvUTF32BE +#define wxMBConvUTF32straight wxMBConvUTF32LE +#endif + + +WXDLLIMPEXP_DATA_BASE(wxMBConvUTF32LE) wxConvUTF32LE; +WXDLLIMPEXP_DATA_BASE(wxMBConvUTF32BE) wxConvUTF32BE; + + +#ifdef WC_UTF16 + +// copy 32bit MB to 16bit String +size_t wxMBConvUTF32straight::MB2WC(wchar_t *buf, const char *psz, size_t n) const +{ + size_t len=0; + + while (*(wxUint32*)psz && (!buf || len < n)) + { + wxUint16 cc[2]; + + size_t pa=encode_utf16(*(wxUint32*)psz, cc); + if (pa == (size_t)-1) + return pa; + + if (buf) + { + *buf++ = cc[0]; + if (pa > 1) + *buf++ = cc[1]; + } + len += pa; + psz += sizeof(wxUint32); + } + if (buf && len 1) + *buf++ = cc[1]; + } + len += pa; + psz += sizeof(wxUint32); + } + + if (buf && lenMB2WC(buf,psz,n) : (size_t)-1; } + return len; +} - size_t WC2MB(char *buf, const wchar_t *psz, size_t n) - { return work ? work->WC2MB(buf,psz,n) : (size_t)-1; } - bool usable() const - { return work!=NULL; } -public: - wxMBConv*work; -}; +#endif // WC_UTF16 // ============================================================================ @@ -563,19 +862,19 @@ public: #define ICONV_CHAR_CAST(x) ((ICONV_CONST char **)(x)) // ---------------------------------------------------------------------------- -// IC_CharSet: encapsulates an iconv character set +// wxMBConv_iconv: encapsulates an iconv character set // ---------------------------------------------------------------------------- -class IC_CharSet : public wxCharacterSet +class wxMBConv_iconv : public wxMBConv { public: - IC_CharSet(const wxChar *name); - virtual ~IC_CharSet(); + wxMBConv_iconv(const wxChar *name); + virtual ~wxMBConv_iconv(); - virtual size_t MB2WC(wchar_t *buf, const char *psz, size_t n); - virtual size_t WC2MB(char *buf, const wchar_t *psz, size_t n); + 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; - bool usable() const + bool IsOk() const { return (m2w != (iconv_t)-1) && (w2m != (iconv_t)-1); } protected: @@ -585,7 +884,7 @@ protected: w2m; private: - // the name (for iconv_open()) of a wide char charset - if none is + // the name (for iconv_open()) of a wide char charset -- if none is // available on this machine, it will remain NULL static const char *ms_wcCharsetName; @@ -594,33 +893,37 @@ private: static bool ms_wcNeedsSwap; }; -const char *IC_CharSet::ms_wcCharsetName = NULL; -bool IC_CharSet::ms_wcNeedsSwap = FALSE; +const char *wxMBConv_iconv::ms_wcCharsetName = NULL; +bool wxMBConv_iconv::ms_wcNeedsSwap = false; -IC_CharSet::IC_CharSet(const wxChar *name) - : wxCharacterSet(name) +wxMBConv_iconv::wxMBConv_iconv(const wxChar *name) { + // Do it the hard way + char cname[100]; + for (size_t i = 0; i < wxStrlen(name)+1; i++) + cname[i] = (char) name[i]; + // check for charset that represents wchar_t: if (ms_wcCharsetName == NULL) { - ms_wcNeedsSwap = FALSE; + ms_wcNeedsSwap = false; // try charset with explicit bytesex info (e.g. "UCS-4LE"): ms_wcCharsetName = WC_NAME_BEST; - m2w = iconv_open(ms_wcCharsetName, wxConvLibc.cWX2MB(name)); + m2w = iconv_open(ms_wcCharsetName, cname); if (m2w == (iconv_t)-1) { // try charset w/o bytesex info (e.g. "UCS4") // and check for bytesex ourselves: ms_wcCharsetName = WC_NAME; - m2w = iconv_open(ms_wcCharsetName, wxConvLibc.cWX2MB(name)); + m2w = iconv_open(ms_wcCharsetName, cname); // last bet, try if it knows WCHAR_T pseudo-charset if (m2w == (iconv_t)-1) { ms_wcCharsetName = "WCHAR_T"; - m2w = iconv_open(ms_wcCharsetName, wxConvLibc.cWX2MB(name)); + m2w = iconv_open(ms_wcCharsetName, cname); } if (m2w != (iconv_t)-1) @@ -645,7 +948,7 @@ IC_CharSet::IC_CharSet(const wxChar *name) { ms_wcCharsetName = NULL; wxLogLastError(wxT("iconv")); - wxLogError(_("Convertion to charset '%s' doesn't work."), name); + wxLogError(_("Conversion to charset '%s' doesn't work."), name); } else { @@ -655,8 +958,8 @@ IC_CharSet::IC_CharSet(const wxChar *name) else { ms_wcCharsetName = NULL; - - // VS: we must not output an error here, since wxWindows will safely + + // VS: we must not output an error here, since wxWidgets will safely // fall back to using wxEncodingConverter. wxLogTrace(wxT("strconv"), wxT("Impossible to convert to/from charset '%s' with iconv, falling back to wxEncodingConverter."), name); //wxLogError( @@ -666,13 +969,13 @@ IC_CharSet::IC_CharSet(const wxChar *name) } else // we already have ms_wcCharsetName { - m2w = iconv_open(ms_wcCharsetName, wxConvLibc.cWX2MB(name)); + m2w = iconv_open(ms_wcCharsetName, cname); } // NB: don't ever pass NULL to iconv_open(), it may crash! if ( ms_wcCharsetName ) { - w2m = iconv_open(wxConvLibc.cWX2MB(name), ms_wcCharsetName); + w2m = iconv_open( cname, ms_wcCharsetName); } else { @@ -680,7 +983,7 @@ IC_CharSet::IC_CharSet(const wxChar *name) } } -IC_CharSet::~IC_CharSet() +wxMBConv_iconv::~wxMBConv_iconv() { if ( m2w != (iconv_t)-1 ) iconv_close(m2w); @@ -688,7 +991,7 @@ IC_CharSet::~IC_CharSet() iconv_close(w2m); } -size_t IC_CharSet::MB2WC(wchar_t *buf, const char *psz, size_t n) +size_t wxMBConv_iconv::MB2WC(wchar_t *buf, const char *psz, size_t n) const { size_t inbuf = strlen(psz); size_t outbuf = n * SIZEOF_WCHAR_T; @@ -710,6 +1013,12 @@ size_t IC_CharSet::MB2WC(wchar_t *buf, const char *psz, size_t n) // convert to native endianness WC_BSWAP(buf /* _not_ bufPtr */, res) } + + // NB: iconv was given only strlen(psz) characters on input, and so + // it couldn't convert the trailing zero. Let's do it ourselves + // if there's some room left for it in the output buffer. + if (res < n) + buf[res] = 0; } else { @@ -739,13 +1048,9 @@ size_t IC_CharSet::MB2WC(wchar_t *buf, const char *psz, size_t n) return res; } -size_t IC_CharSet::WC2MB(char *buf, const wchar_t *psz, size_t n) +size_t wxMBConv_iconv::WC2MB(char *buf, const wchar_t *psz, size_t n) const { -#if defined(__BORLANDC__) && (__BORLANDC__ > 0x530) - size_t inbuf = std::wcslen(psz) * SIZEOF_WCHAR_T; -#else - size_t inbuf = ::wcslen(psz) * SIZEOF_WCHAR_T; -#endif + size_t inbuf = wxWcslen(psz) * SIZEOF_WCHAR_T; size_t outbuf = n; size_t res, cres; @@ -769,6 +1074,12 @@ size_t IC_CharSet::WC2MB(char *buf, const wchar_t *psz, size_t n) cres = iconv( w2m, ICONV_CHAR_CAST(&psz), &inbuf, &buf, &outbuf ); res = n-outbuf; + + // NB: iconv was given only wcslen(psz) characters on input, and so + // it couldn't convert the trailing zero. Let's do it ourselves + // if there's some room left for it in the output buffer. + if (res < n) + buf[0] = 0; } else { @@ -802,45 +1113,314 @@ size_t IC_CharSet::WC2MB(char *buf, const wchar_t *psz, size_t n) #endif // HAVE_ICONV + // ============================================================================ // Win32 conversion classes // ============================================================================ -#if defined(__WIN32__) && !defined(__WXMICROWIN__) -class CP_CharSet : public wxCharacterSet +#ifdef wxHAVE_WIN32_MB2WC + +// from utils.cpp +#if wxUSE_FONTMAP +extern WXDLLIMPEXP_BASE long wxCharsetToCodepage(const wxChar *charset); +extern WXDLLIMPEXP_BASE long wxEncodingToCodepage(wxFontEncoding encoding); +#endif + +class wxMBConv_win32 : public wxMBConv { public: - CP_CharSet(const wxChar* name) - : wxCharacterSet(name) + wxMBConv_win32() + { + m_CodePage = CP_ACP; + } + +#if wxUSE_FONTMAP + wxMBConv_win32(const wxChar* name) + { + m_CodePage = wxCharsetToCodepage(name); + } + + wxMBConv_win32(wxFontEncoding encoding) + { + m_CodePage = wxEncodingToCodepage(encoding); + } +#endif + + size_t MB2WC(wchar_t *buf, const char *psz, size_t n) const + { + const size_t len = ::MultiByteToWideChar + ( + m_CodePage, // code page + 0, // flags (none) + psz, // input string + -1, // its length (NUL-terminated) + buf, // output string + buf ? n : 0 // size of output buffer + ); + + // note that it returns count of written chars for buf != NULL and size + // of the needed buffer for buf == NULL so in either case the length of + // the string (which never includes the terminating NUL) is one less + return len ? len - 1 : (size_t)-1; + } + + size_t WC2MB(char *buf, const wchar_t *pwz, size_t n) const + { + /* + we have a problem here: by default, WideCharToMultiByte() may + replace characters unrepresentable in the target code page with bad + quality approximations such as turning "1/2" symbol (U+00BD) into + "1" for the code pages which don't have it and we, obviously, want + to avoid this at any price + + the trouble is that this function does it _silently_, i.e. it won't + even tell us whether it did or not... Win98/2000 and higher provide + WC_NO_BEST_FIT_CHARS but it doesn't work for the older systems and + we have to resort to a round trip, i.e. check that converting back + results in the same string -- this is, of course, expensive but + otherwise we simply can't be sure to not garble the data. + */ + + // determine if we can rely on WC_NO_BEST_FIT_CHARS: according to MSDN + // it doesn't work with CJK encodings (which we test for rather roughly + // here...) nor with UTF-7/8 nor, of course, with Windows versions not + // supporting it + BOOL usedDef wxDUMMY_INITIALIZE(false), + *pUsedDef; + int flags; + if ( CanUseNoBestFit() && m_CodePage < 50000 ) + { + // it's our lucky day + flags = WC_NO_BEST_FIT_CHARS; + pUsedDef = &usedDef; + } + else // old system or unsupported encoding { - m_CodePage = wxCharsetToCodepage(name); + flags = 0; + pUsedDef = NULL; } - size_t MB2WC(wchar_t *buf, const char *psz, size_t n) - { - size_t len = - MultiByteToWideChar(m_CodePage, 0, psz, -1, buf, buf ? n : 0); - //VS: returns # of written chars for buf!=NULL and *size* - // needed buffer for buf==NULL - return len ? (buf ? len : len-1) : (size_t)-1; + const size_t len = ::WideCharToMultiByte + ( + m_CodePage, // code page + flags, // either none or no best fit + pwz, // input string + -1, // it is (wide) NUL-terminated + buf, // output buffer + buf ? n : 0, // and its size + NULL, // default "replacement" char + pUsedDef // [out] was it used? + ); + + if ( !len ) + { + // function totally failed + return (size_t)-1; + } + + // if we were really converting, check if we succeeded + if ( buf ) + { + if ( flags ) + { + // check if the conversion failed, i.e. if any replacements + // were done + if ( usedDef ) + return (size_t)-1; + } + else // we must resort to double tripping... + { + wxWCharBuffer wcBuf(n); + if ( MB2WC(wcBuf.data(), buf, n) == (size_t)-1 || + wcscmp(wcBuf, pwz) != 0 ) + { + // we didn't obtain the same thing we started from, hence + // the conversion was lossy and we consider that it failed + return (size_t)-1; + } + } + } + + // see the comment above for the reason of "len - 1" + return len - 1; } - size_t WC2MB(char *buf, const wchar_t *psz, size_t n) + bool IsOk() const { return m_CodePage != -1; } + +private: + static bool CanUseNoBestFit() { - size_t len = WideCharToMultiByte(m_CodePage, 0, psz, -1, buf, - buf ? n : 0, NULL, NULL); - //VS: returns # of written chars for buf!=NULL and *size* - // needed buffer for buf==NULL - return len ? (buf ? len : len-1) : (size_t)-1; + static int s_isWin98Or2k = -1; + + if ( s_isWin98Or2k == -1 ) + { + int verMaj, verMin; + switch ( wxGetOsVersion(&verMaj, &verMin) ) + { + case wxWIN95: + s_isWin98Or2k = verMaj >= 4 && verMin >= 10; + break; + + case wxWINDOWS_NT: + s_isWin98Or2k = verMaj >= 5; + break; + + default: + // unknown, be conseravtive by default + s_isWin98Or2k = 0; + } + + wxASSERT_MSG( s_isWin98Or2k != -1, _T("should be set above") ); + } + + return s_isWin98Or2k == 1; } - bool usable() const - { return m_CodePage != -1; } + long m_CodePage; +}; + +#endif // wxHAVE_WIN32_MB2WC + +// ============================================================================ +// Mac conversion classes +// ============================================================================ + +#if defined(__WXMAC__) && defined(TARGET_CARBON) +class wxMBConv_mac : public wxMBConv +{ public: - long m_CodePage; + wxMBConv_mac() + { + Init(CFStringGetSystemEncoding()) ; + } + + wxMBConv_mac(const wxChar* name) + { + Init( wxMacGetSystemEncFromFontEnc(wxFontMapper::Get()->CharsetToEncoding(name, FALSE) ) ) ; + } + + wxMBConv_mac(wxFontEncoding encoding) + { + Init( wxMacGetSystemEncFromFontEnc(encoding) ); + } + + ~wxMBConv_mac() + { + OSStatus status = noErr ; + status = TECDisposeConverter(m_MB2WC_converter); + status = TECDisposeConverter(m_WC2MB_converter); + } + + + void Init( TextEncodingBase encoding) + { + OSStatus status = noErr ; + m_char_encoding = encoding ; + m_unicode_encoding = CreateTextEncoding(kTextEncodingUnicodeDefault,0,kUnicode16BitFormat) ; + + status = TECCreateConverter(&m_MB2WC_converter, + m_char_encoding, + m_unicode_encoding); + status = TECCreateConverter(&m_WC2MB_converter, + m_unicode_encoding, + m_char_encoding); + } + + size_t MB2WC(wchar_t *buf, const char *psz, size_t n) const + { + OSStatus status = noErr ; + ByteCount byteOutLen ; + ByteCount byteInLen = strlen(psz) ; + wchar_t *tbuf = NULL ; + UniChar* ubuf = NULL ; + size_t res = 0 ; + + if (buf == NULL) + { + n = byteInLen ; + tbuf = (wchar_t*) malloc( n * SIZEOF_WCHAR_T) ; + } + ByteCount byteBufferLen = n * sizeof( UniChar ) ; +#if SIZEOF_WCHAR_T == 4 + ubuf = (UniChar*) malloc( byteBufferLen + 2 ) ; +#else + ubuf = (UniChar*) (buf ? buf : tbuf) ; +#endif + status = TECConvertText(m_MB2WC_converter, (ConstTextPtr) psz , byteInLen, &byteInLen, + (TextPtr) ubuf , byteBufferLen, &byteOutLen); +#if SIZEOF_WCHAR_T == 4 + // we have to terminate here, because n might be larger for the trailing zero, and if UniChar + // is not properly terminated we get random characters at the end + ubuf[byteOutLen / sizeof( UniChar ) ] = 0 ; + wxMBConvUTF16BE converter ; + res = converter.MB2WC( (buf ? buf : tbuf) , (const char*)ubuf , n ) ; + free( ubuf ) ; +#else + res = byteOutLen / sizeof( UniChar ) ; +#endif + if ( buf == NULL ) + free(tbuf) ; + + if ( buf && res < n) + buf[res] = 0; + + return res ; + } + + size_t WC2MB(char *buf, const wchar_t *psz, size_t n) const + { + OSStatus status = noErr ; + ByteCount byteOutLen ; + ByteCount byteInLen = wxWcslen(psz) * SIZEOF_WCHAR_T ; + + char *tbuf = NULL ; + + if (buf == NULL) + { + // worst case + n = byteInLen * 2 ; + tbuf = (char*) malloc( n ) ; + } + + ByteCount byteBufferLen = n ; + UniChar* ubuf = NULL ; +#if SIZEOF_WCHAR_T == 4 + wxMBConvUTF16BE converter ; + size_t unicharlen = converter.WC2MB( NULL , psz , 0 ) ; + byteInLen = unicharlen ; + ubuf = (UniChar*) malloc( byteInLen + 2 ) ; + converter.WC2MB( (char*) ubuf , psz, unicharlen + 2 ) ; +#else + ubuf = (UniChar*) psz ; +#endif + status = TECConvertText(m_WC2MB_converter, (ConstTextPtr) ubuf , byteInLen, &byteInLen, + (TextPtr) (buf ? buf : tbuf) , byteBufferLen, &byteOutLen); +#if SIZEOF_WCHAR_T == 4 + free( ubuf ) ; +#endif + if ( buf == NULL ) + free(tbuf) ; + + size_t res = byteOutLen ; + if ( buf && res < n) + buf[res] = 0; + + return res ; + } + + bool IsOk() const + { return m_MB2WC_converter != NULL && m_WC2MB_converter != NULL ; } + +private: + TECObjectRef m_MB2WC_converter ; + TECObjectRef m_WC2MB_converter ; + + TextEncodingBase m_char_encoding ; + TextEncodingBase m_unicode_encoding ; }; -#endif // __WIN32__ + +#endif // defined(__WXMAC__) && defined(TARGET_CARBON) // ============================================================================ // wxEncodingConverter based conversion classes @@ -848,21 +1428,36 @@ public: #if wxUSE_FONTMAP -class EC_CharSet : public wxCharacterSet +class wxMBConv_wxwin : public wxMBConv { +private: + void Init() + { + m_ok = m2w.Init(m_enc, wxFONTENCODING_UNICODE) && + w2m.Init(wxFONTENCODING_UNICODE, m_enc); + } + public: // temporarily just use wxEncodingConverter stuff, // so that it works while a better implementation is built - EC_CharSet(const wxChar* name) : wxCharacterSet(name), - enc(wxFONTENCODING_SYSTEM) + wxMBConv_wxwin(const wxChar* name) { if (name) - enc = wxTheFontMapper->CharsetToEncoding(name, FALSE); - m2w.Init(enc, wxFONTENCODING_UNICODE); - w2m.Init(wxFONTENCODING_UNICODE, enc); + m_enc = wxFontMapper::Get()->CharsetToEncoding(name, false); + else + m_enc = wxFONTENCODING_SYSTEM; + + Init(); + } + + wxMBConv_wxwin(wxFontEncoding enc) + { + m_enc = enc; + + Init(); } - size_t MB2WC(wchar_t *buf, const char *psz, size_t n) + size_t MB2WC(wchar_t *buf, const char *psz, size_t WXUNUSED(n)) const { size_t inbuf = strlen(psz); if (buf) @@ -870,97 +1465,97 @@ public: return inbuf; } - size_t WC2MB(char *buf, const wchar_t *psz, size_t n) + size_t WC2MB(char *buf, const wchar_t *psz, size_t WXUNUSED(n)) const { -#if ( defined(__BORLANDC__) && (__BORLANDC__ > 0x530) ) \ - || ( defined(__MWERKS__) && defined(__WXMSW__) ) - size_t inbuf = std::wcslen(psz); -#else - size_t inbuf = ::wcslen(psz); -#endif + const size_t inbuf = wxWcslen(psz); if (buf) w2m.Convert(psz,buf); return inbuf; } - bool usable() const - { return (enc!=wxFONTENCODING_SYSTEM) && (enc!=wxFONTENCODING_DEFAULT); } + bool IsOk() const { return m_ok; } public: - wxFontEncoding enc; + wxFontEncoding m_enc; wxEncodingConverter m2w, w2m; + + // were we initialized successfully? + bool m_ok; + + DECLARE_NO_COPY_CLASS(wxMBConv_wxwin) }; #endif // wxUSE_FONTMAP -// ---------------------------------------------------------------------------- -// the function creating the wxCharacterSet for the specified charset on the -// current system, trying all possibilities -// ---------------------------------------------------------------------------- +// ============================================================================ +// wxCSConv implementation +// ============================================================================ + +void wxCSConv::Init() +{ + m_name = NULL; + m_convReal = NULL; + m_deferred = true; +} -static wxCharacterSet *wxGetCharacterSet(const wxChar *name) +wxCSConv::wxCSConv(const wxChar *charset) { - wxCharacterSet *cset = NULL; - if (name) + Init(); + + if ( charset ) { - if (wxStricmp(name, wxT("UTF8")) == 0 || wxStricmp(name, wxT("UTF-8")) == 0) - { - cset = new ID_CharSet(name, &wxConvUTF8); - } - else - { -#ifdef HAVE_ICONV - cset = new IC_CharSet(name); // may not take NULL -#endif - } + SetName(charset); } - if (cset && cset->usable()) - return cset; + m_encoding = wxFONTENCODING_SYSTEM; +} - if (cset) +wxCSConv::wxCSConv(wxFontEncoding encoding) +{ + if ( encoding == wxFONTENCODING_MAX || encoding == wxFONTENCODING_DEFAULT ) { - delete cset; - cset = NULL; - } + wxFAIL_MSG( _T("invalid encoding value in wxCSConv ctor") ); -#if defined(__WIN32__) && !defined(__WXMICROWIN__) - cset = new CP_CharSet(name); // may take NULL - if (cset->usable()) - return cset; + encoding = wxFONTENCODING_SYSTEM; + } - delete cset; -#endif // __WIN32__ + Init(); -#if wxUSE_FONTMAP - cset = new EC_CharSet(name); - if (cset->usable()) - return cset; -#endif // wxUSE_FONTMAP + m_encoding = encoding; +} - delete cset; - wxLogError(_("Unknown encoding '%s'!"), name); - return NULL; +wxCSConv::~wxCSConv() +{ + Clear(); } -// ============================================================================ -// wxCSConv implementation -// ============================================================================ +wxCSConv::wxCSConv(const wxCSConv& conv) + : wxMBConv() +{ + Init(); -wxCSConv::wxCSConv(const wxChar *charset) + SetName(conv.m_name); + m_encoding = conv.m_encoding; +} + +wxCSConv& wxCSConv::operator=(const wxCSConv& conv) { - m_name = (wxChar *)NULL; - m_cset = (wxCharacterSet *) NULL; - m_deferred = TRUE; + Clear(); - SetName(charset); + SetName(conv.m_name); + m_encoding = conv.m_encoding; + + return *this; } -wxCSConv::~wxCSConv() +void wxCSConv::Clear() { free(m_name); - delete m_cset; + delete m_convReal; + + m_name = NULL; + m_convReal = NULL; } void wxCSConv::SetName(const wxChar *charset) @@ -968,33 +1563,179 @@ void wxCSConv::SetName(const wxChar *charset) if (charset) { m_name = wxStrdup(charset); - m_deferred = TRUE; + m_deferred = true; + } +} + +wxMBConv *wxCSConv::DoCreate() const +{ + // check for the special case of ASCII or ISO8859-1 charset: as we have + // special knowledge of it anyhow, we don't need to create a special + // conversion object + if ( m_encoding == wxFONTENCODING_ISO8859_1 ) + { + // don't convert at all + return NULL; + } + + // we trust OS to do conversion better than we can so try external + // conversion methods first + // + // the full order is: + // 1. OS conversion (iconv() under Unix or Win32 API) + // 2. hard coded conversions for UTF + // 3. wxEncodingConverter as fall back + + // step (1) +#ifdef HAVE_ICONV +#if !wxUSE_FONTMAP + if ( m_name ) +#endif // !wxUSE_FONTMAP + { + wxString name(m_name); + +#if wxUSE_FONTMAP + if ( name.empty() ) + name = wxFontMapper::Get()->GetEncodingName(m_encoding); +#endif // wxUSE_FONTMAP + + wxMBConv_iconv *conv = new wxMBConv_iconv(name); + if ( conv->IsOk() ) + return conv; + + delete conv; + } +#endif // HAVE_ICONV + +#ifdef wxHAVE_WIN32_MB2WC + { +#if wxUSE_FONTMAP + wxMBConv_win32 *conv = m_name ? new wxMBConv_win32(m_name) + : new wxMBConv_win32(m_encoding); + if ( conv->IsOk() ) + return conv; + + delete conv; +#else + return NULL; +#endif + } +#endif // wxHAVE_WIN32_MB2WC +#if defined(__WXMAC__) + { + if ( m_name || ( m_encoding < wxFONTENCODING_UTF16BE ) ) + { + + wxMBConv_mac *conv = m_name ? new wxMBConv_mac(m_name) + : new wxMBConv_mac(m_encoding); + if ( conv->IsOk() ) + return conv; + + delete conv; + } + } +#endif + // step (2) + wxFontEncoding enc = m_encoding; +#if wxUSE_FONTMAP + if ( enc == wxFONTENCODING_SYSTEM && m_name ) + { + // use "false" to suppress interactive dialogs -- we can be called from + // anywhere and popping up a dialog from here is the last thing we want to + // do + enc = wxFontMapper::Get()->CharsetToEncoding(m_name, false); + } +#endif // wxUSE_FONTMAP + + switch ( enc ) + { + case wxFONTENCODING_UTF7: + return new wxMBConvUTF7; + + case wxFONTENCODING_UTF8: + return new wxMBConvUTF8; + + case wxFONTENCODING_UTF16BE: + return new wxMBConvUTF16BE; + + case wxFONTENCODING_UTF16LE: + return new wxMBConvUTF16LE; + + case wxFONTENCODING_UTF32BE: + return new wxMBConvUTF32BE; + + case wxFONTENCODING_UTF32LE: + return new wxMBConvUTF32LE; + + default: + // nothing to do but put here to suppress gcc warnings + ; + } + + // step (3) +#if wxUSE_FONTMAP + { + wxMBConv_wxwin *conv = m_name ? new wxMBConv_wxwin(m_name) + : new wxMBConv_wxwin(m_encoding); + if ( conv->IsOk() ) + return conv; + + delete conv; + } +#endif // wxUSE_FONTMAP + + // NB: This is a hack to prevent deadlock. What could otherwise happen + // in Unicode build: wxConvLocal creation ends up being here + // because of some failure and logs the error. But wxLog will try to + // attach timestamp, for which it will need wxConvLocal (to convert + // time to char* and then wchar_t*), but that fails, tries to log + // error, but wxLog has a (already locked) critical section that + // guards static buffer. + static bool alreadyLoggingError = false; + if (!alreadyLoggingError) + { + alreadyLoggingError = true; + wxLogError(_("Cannot convert from the charset '%s'!"), + m_name ? m_name + : +#if wxUSE_FONTMAP + wxFontMapper::GetEncodingDescription(m_encoding).c_str() +#else // !wxUSE_FONTMAP + wxString::Format(_("encoding %s"), m_encoding).c_str() +#endif // wxUSE_FONTMAP/!wxUSE_FONTMAP + ); + alreadyLoggingError = false; } + + return NULL; } -void wxCSConv::LoadNow() +void wxCSConv::CreateConvIfNeeded() const { - if (m_deferred) + if ( m_deferred ) { - if ( !m_name ) + wxCSConv *self = (wxCSConv *)this; // const_cast + +#if wxUSE_INTL + // if we don't have neither the name nor the encoding, use the default + // encoding for this system + if ( !m_name && m_encoding == wxFONTENCODING_SYSTEM ) { - wxString name = wxLocale::GetSystemEncodingName(); - if ( !name.empty() ) - SetName(name); + self->m_name = wxStrdup(wxLocale::GetSystemEncodingName()); } +#endif // wxUSE_INTL - // wxGetCharacterSet() complains about NULL name - m_cset = m_name ? wxGetCharacterSet(m_name) : NULL; - m_deferred = FALSE; + self->m_convReal = DoCreate(); + self->m_deferred = false; } } size_t wxCSConv::MB2WC(wchar_t *buf, const char *psz, size_t n) const { - ((wxCSConv *)this)->LoadNow(); // discard constness + CreateConvIfNeeded(); - if (m_cset) - return m_cset->MB2WC(buf, psz, n); + if (m_convReal) + return m_convReal->MB2WC(buf, psz, n); // latin-1 (direct) size_t len = strlen(psz); @@ -1010,35 +1751,65 @@ size_t wxCSConv::MB2WC(wchar_t *buf, const char *psz, size_t n) const size_t wxCSConv::WC2MB(char *buf, const wchar_t *psz, size_t n) const { - ((wxCSConv *)this)->LoadNow(); // discard constness + CreateConvIfNeeded(); - if (m_cset) - return m_cset->WC2MB(buf, psz, n); + if (m_convReal) + return m_convReal->WC2MB(buf, psz, n); // latin-1 (direct) -#if ( defined(__BORLANDC__) && (__BORLANDC__ > 0x530) ) \ - || ( defined(__MWERKS__) && defined(__WXMSW__) ) - size_t len=std::wcslen(psz); -#else - size_t len=::wcslen(psz); -#endif + const size_t len = wxWcslen(psz); if (buf) { for (size_t c = 0; c <= len; c++) - buf[c] = (psz[c] > 0xff) ? '?' : psz[c]; + { + if (psz[c] > 0xFF) + return (size_t)-1; + buf[c] = psz[c]; + } + } + else + { + for (size_t c = 0; c <= len; c++) + { + if (psz[c] > 0xFF) + return (size_t)-1; + } } return len; } -#else // !wxUSE_WCHAR_T - // ---------------------------------------------------------------------------- -// stand-ins in absence of wchar_t +// globals // ---------------------------------------------------------------------------- -WXDLLEXPORT_DATA(wxMBConv) wxConvLibc, wxConvFile; +#ifdef __WINDOWS__ + static wxMBConv_win32 wxConvLibcObj; +#else + static wxMBConvLibc wxConvLibcObj; +#endif + +static wxCSConv wxConvLocalObj(wxFONTENCODING_SYSTEM); +static wxCSConv wxConvISO8859_1Obj(wxFONTENCODING_ISO8859_1); +static wxMBConvUTF7 wxConvUTF7Obj; +static wxMBConvUTF8 wxConvUTF8Obj; + + +WXDLLIMPEXP_DATA_BASE(wxMBConv&) wxConvLibc = wxConvLibcObj; +WXDLLIMPEXP_DATA_BASE(wxCSConv&) wxConvLocal = wxConvLocalObj; +WXDLLIMPEXP_DATA_BASE(wxCSConv&) wxConvISO8859_1 = wxConvISO8859_1Obj; +WXDLLIMPEXP_DATA_BASE(wxMBConvUTF7&) wxConvUTF7 = wxConvUTF7Obj; +WXDLLIMPEXP_DATA_BASE(wxMBConvUTF8&) wxConvUTF8 = wxConvUTF8Obj; +WXDLLIMPEXP_DATA_BASE(wxMBConv *) wxConvCurrent = &wxConvLibcObj; + +#else // !wxUSE_WCHAR_T + +// stand-ins in absence of wchar_t +WXDLLIMPEXP_DATA_BASE(wxMBConv) wxConvLibc, + wxConvISO8859_1, + wxConvLocal, + wxConvUTF8; -#endif // wxUSE_WCHAR_T +#endif // wxUSE_WCHAR_T/!wxUSE_WCHAR_T