+ 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
+ {
+ // no destination buffer... convert using temp buffer
+ // to calculate destination buffer requirement
+ char tbuf[16];
+ res = 0;
+ do {
+ buf = tbuf; outbuf = 16;
+
+ cres = iconv( w2m, ICONV_CHAR_CAST(&psz), &inbuf, &buf, &outbuf );
+
+ res += 16 - outbuf;
+ } while ((cres==(size_t)-1) && (errno==E2BIG));
+ }
+
+ if (ms_wcNeedsSwap)
+ {
+ free(tmpbuf);
+ }
+
+ if (ICONV_FAILED(cres, inbuf))
+ {
+ //VS: it is ok if iconv fails, hence trace only
+ wxLogTrace(wxT("strconv"), wxT("iconv failed: %s"), wxSysErrorMsg(wxSysErrorCode()));
+ return (size_t)-1;
+ }
+
+ return res;
+}
+
+#endif // HAVE_ICONV
+
+
+// ============================================================================
+// Win32 conversion classes
+// ============================================================================
+
+#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:
+ 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
+ {
+ // note that we have to use MB_ERR_INVALID_CHARS flag as it without it
+ // the behaviour is not compatible with the Unix version (using iconv)
+ // and break the library itself, e.g. wxTextInputStream::NextChar()
+ // wouldn't work if reading an incomplete MB char didn't result in an
+ // error
+ //
+ // note however that using MB_ERR_INVALID_CHARS with CP_UTF7 results in
+ // an error (tested under Windows Server 2003) and apparently it is
+ // done on purpose, i.e. the function accepts any input in this case
+ // and although I'd prefer to return error on ill-formed output, our
+ // own wxMBConvUTF7 doesn't detect errors (e.g. lone "+" which is
+ // explicitly ill-formed according to RFC 2152) neither so we don't
+ // even have any fallback here...
+ int flags = m_CodePage == CP_UTF7 ? 0 : MB_ERR_INVALID_CHARS;
+
+ const size_t len = ::MultiByteToWideChar
+ (
+ m_CodePage, // code page
+ flags, // flags: fall on error
+ 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);
+ BOOL *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
+ {
+ flags = 0;
+ pUsedDef = NULL;
+ }
+
+ 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;
+ }
+
+ bool IsOk() const { return m_CodePage != -1; }
+
+private:
+ static bool CanUseNoBestFit()
+ {
+ 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;
+ }
+
+ long m_CodePage;
+};
+
+#endif // wxHAVE_WIN32_MB2WC
+
+// ============================================================================
+// Cocoa conversion classes
+// ============================================================================
+
+#if defined(__WXCOCOA__)
+
+// RN: There is no UTF-32 support in either Core Foundation or
+// Cocoa. Strangely enough, internally Core Foundation uses
+// UTF 32 internally quite a bit - its just not public (yet).
+
+#include <CoreFoundation/CFString.h>
+#include <CoreFoundation/CFStringEncodingExt.h>
+
+CFStringEncoding wxCFStringEncFromFontEnc(wxFontEncoding encoding)
+{
+ CFStringEncoding enc = kCFStringEncodingInvalidId ;
+ if ( encoding == wxFONTENCODING_DEFAULT )
+ {
+ enc = CFStringGetSystemEncoding();
+ }
+ else switch( encoding)
+ {
+ case wxFONTENCODING_ISO8859_1 :
+ enc = kCFStringEncodingISOLatin1 ;
+ break ;
+ case wxFONTENCODING_ISO8859_2 :
+ enc = kCFStringEncodingISOLatin2;
+ break ;
+ case wxFONTENCODING_ISO8859_3 :
+ enc = kCFStringEncodingISOLatin3 ;
+ break ;
+ case wxFONTENCODING_ISO8859_4 :
+ enc = kCFStringEncodingISOLatin4;
+ break ;
+ case wxFONTENCODING_ISO8859_5 :
+ enc = kCFStringEncodingISOLatinCyrillic;
+ break ;
+ case wxFONTENCODING_ISO8859_6 :
+ enc = kCFStringEncodingISOLatinArabic;
+ break ;
+ case wxFONTENCODING_ISO8859_7 :
+ enc = kCFStringEncodingISOLatinGreek;
+ break ;
+ case wxFONTENCODING_ISO8859_8 :
+ enc = kCFStringEncodingISOLatinHebrew;
+ break ;
+ case wxFONTENCODING_ISO8859_9 :
+ enc = kCFStringEncodingISOLatin5;
+ break ;
+ case wxFONTENCODING_ISO8859_10 :
+ enc = kCFStringEncodingISOLatin6;
+ break ;
+ case wxFONTENCODING_ISO8859_11 :
+ enc = kCFStringEncodingISOLatinThai;
+ break ;
+ case wxFONTENCODING_ISO8859_13 :
+ enc = kCFStringEncodingISOLatin7;
+ break ;
+ case wxFONTENCODING_ISO8859_14 :
+ enc = kCFStringEncodingISOLatin8;
+ break ;
+ case wxFONTENCODING_ISO8859_15 :
+ enc = kCFStringEncodingISOLatin9;
+ break ;
+
+ case wxFONTENCODING_KOI8 :
+ enc = kCFStringEncodingKOI8_R;
+ break ;
+ case wxFONTENCODING_ALTERNATIVE : // MS-DOS CP866
+ enc = kCFStringEncodingDOSRussian;
+ break ;
+
+// case wxFONTENCODING_BULGARIAN :
+// enc = ;
+// break ;
+
+ case wxFONTENCODING_CP437 :
+ enc =kCFStringEncodingDOSLatinUS ;
+ break ;
+ case wxFONTENCODING_CP850 :
+ enc = kCFStringEncodingDOSLatin1;
+ break ;
+ case wxFONTENCODING_CP852 :
+ enc = kCFStringEncodingDOSLatin2;
+ break ;
+ case wxFONTENCODING_CP855 :
+ enc = kCFStringEncodingDOSCyrillic;
+ break ;
+ case wxFONTENCODING_CP866 :
+ enc =kCFStringEncodingDOSRussian ;
+ break ;
+ case wxFONTENCODING_CP874 :
+ enc = kCFStringEncodingDOSThai;
+ break ;
+ case wxFONTENCODING_CP932 :
+ enc = kCFStringEncodingDOSJapanese;
+ break ;
+ case wxFONTENCODING_CP936 :
+ enc =kCFStringEncodingDOSChineseSimplif ;
+ break ;
+ case wxFONTENCODING_CP949 :
+ enc = kCFStringEncodingDOSKorean;
+ break ;
+ case wxFONTENCODING_CP950 :
+ enc = kCFStringEncodingDOSChineseTrad;
+ break ;
+ case wxFONTENCODING_CP1250 :
+ enc = kCFStringEncodingWindowsLatin2;
+ break ;
+ case wxFONTENCODING_CP1251 :
+ enc =kCFStringEncodingWindowsCyrillic ;
+ break ;
+ case wxFONTENCODING_CP1252 :
+ enc =kCFStringEncodingWindowsLatin1 ;
+ break ;
+ case wxFONTENCODING_CP1253 :
+ enc = kCFStringEncodingWindowsGreek;
+ break ;
+ case wxFONTENCODING_CP1254 :
+ enc = kCFStringEncodingWindowsLatin5;
+ break ;
+ case wxFONTENCODING_CP1255 :
+ enc =kCFStringEncodingWindowsHebrew ;
+ break ;
+ case wxFONTENCODING_CP1256 :
+ enc =kCFStringEncodingWindowsArabic ;
+ break ;
+ case wxFONTENCODING_CP1257 :
+ enc = kCFStringEncodingWindowsBalticRim;
+ break ;
+// This only really encodes to UTF7 (if that) evidently
+// case wxFONTENCODING_UTF7 :
+// enc = kCFStringEncodingNonLossyASCII ;
+// break ;
+ case wxFONTENCODING_UTF8 :
+ enc = kCFStringEncodingUTF8 ;
+ break ;
+ case wxFONTENCODING_EUC_JP :
+ enc = kCFStringEncodingEUC_JP;
+ break ;
+ case wxFONTENCODING_UTF16 :
+ enc = kCFStringEncodingUnicode ;
+ break ;
+ case wxFONTENCODING_MACROMAN :
+ enc = kCFStringEncodingMacRoman ;
+ break ;
+ case wxFONTENCODING_MACJAPANESE :
+ enc = kCFStringEncodingMacJapanese ;
+ break ;
+ case wxFONTENCODING_MACCHINESETRAD :
+ enc = kCFStringEncodingMacChineseTrad ;
+ break ;
+ case wxFONTENCODING_MACKOREAN :
+ enc = kCFStringEncodingMacKorean ;
+ break ;
+ case wxFONTENCODING_MACARABIC :
+ enc = kCFStringEncodingMacArabic ;
+ break ;
+ case wxFONTENCODING_MACHEBREW :
+ enc = kCFStringEncodingMacHebrew ;
+ break ;
+ case wxFONTENCODING_MACGREEK :
+ enc = kCFStringEncodingMacGreek ;
+ break ;
+ case wxFONTENCODING_MACCYRILLIC :
+ enc = kCFStringEncodingMacCyrillic ;
+ break ;
+ case wxFONTENCODING_MACDEVANAGARI :
+ enc = kCFStringEncodingMacDevanagari ;
+ break ;
+ case wxFONTENCODING_MACGURMUKHI :
+ enc = kCFStringEncodingMacGurmukhi ;
+ break ;
+ case wxFONTENCODING_MACGUJARATI :
+ enc = kCFStringEncodingMacGujarati ;
+ break ;
+ case wxFONTENCODING_MACORIYA :
+ enc = kCFStringEncodingMacOriya ;
+ break ;
+ case wxFONTENCODING_MACBENGALI :
+ enc = kCFStringEncodingMacBengali ;
+ break ;
+ case wxFONTENCODING_MACTAMIL :
+ enc = kCFStringEncodingMacTamil ;
+ break ;
+ case wxFONTENCODING_MACTELUGU :
+ enc = kCFStringEncodingMacTelugu ;
+ break ;
+ case wxFONTENCODING_MACKANNADA :
+ enc = kCFStringEncodingMacKannada ;
+ break ;
+ case wxFONTENCODING_MACMALAJALAM :
+ enc = kCFStringEncodingMacMalayalam ;
+ break ;
+ case wxFONTENCODING_MACSINHALESE :
+ enc = kCFStringEncodingMacSinhalese ;
+ break ;
+ case wxFONTENCODING_MACBURMESE :
+ enc = kCFStringEncodingMacBurmese ;
+ break ;
+ case wxFONTENCODING_MACKHMER :
+ enc = kCFStringEncodingMacKhmer ;
+ break ;
+ case wxFONTENCODING_MACTHAI :
+ enc = kCFStringEncodingMacThai ;
+ break ;
+ case wxFONTENCODING_MACLAOTIAN :
+ enc = kCFStringEncodingMacLaotian ;
+ break ;
+ case wxFONTENCODING_MACGEORGIAN :
+ enc = kCFStringEncodingMacGeorgian ;
+ break ;
+ case wxFONTENCODING_MACARMENIAN :
+ enc = kCFStringEncodingMacArmenian ;
+ break ;
+ case wxFONTENCODING_MACCHINESESIMP :
+ enc = kCFStringEncodingMacChineseSimp ;
+ break ;
+ case wxFONTENCODING_MACTIBETAN :
+ enc = kCFStringEncodingMacTibetan ;
+ break ;
+ case wxFONTENCODING_MACMONGOLIAN :
+ enc = kCFStringEncodingMacMongolian ;
+ break ;
+ case wxFONTENCODING_MACETHIOPIC :
+ enc = kCFStringEncodingMacEthiopic ;
+ break ;
+ case wxFONTENCODING_MACCENTRALEUR :
+ enc = kCFStringEncodingMacCentralEurRoman ;
+ break ;
+ case wxFONTENCODING_MACVIATNAMESE :
+ enc = kCFStringEncodingMacVietnamese ;
+ break ;
+ case wxFONTENCODING_MACARABICEXT :
+ enc = kCFStringEncodingMacExtArabic ;
+ break ;
+ case wxFONTENCODING_MACSYMBOL :
+ enc = kCFStringEncodingMacSymbol ;
+ break ;
+ case wxFONTENCODING_MACDINGBATS :
+ enc = kCFStringEncodingMacDingbats ;
+ break ;
+ case wxFONTENCODING_MACTURKISH :
+ enc = kCFStringEncodingMacTurkish ;
+ break ;
+ case wxFONTENCODING_MACCROATIAN :
+ enc = kCFStringEncodingMacCroatian ;
+ break ;
+ case wxFONTENCODING_MACICELANDIC :
+ enc = kCFStringEncodingMacIcelandic ;
+ break ;
+ case wxFONTENCODING_MACROMANIAN :
+ enc = kCFStringEncodingMacRomanian ;
+ break ;
+ case wxFONTENCODING_MACCELTIC :
+ enc = kCFStringEncodingMacCeltic ;
+ break ;
+ case wxFONTENCODING_MACGAELIC :
+ enc = kCFStringEncodingMacGaelic ;
+ break ;
+// case wxFONTENCODING_MACKEYBOARD :
+// enc = kCFStringEncodingMacKeyboardGlyphs ;
+// break ;
+ default :
+ // because gcc is picky
+ break ;
+ } ;
+ return enc ;
+}
+
+class wxMBConv_cocoa : public wxMBConv
+{
+public:
+ wxMBConv_cocoa()
+ {
+ Init(CFStringGetSystemEncoding()) ;
+ }
+
+#if wxUSE_FONTMAP
+ wxMBConv_cocoa(const wxChar* name)
+ {
+ Init( wxCFStringEncFromFontEnc(wxFontMapperBase::Get()->CharsetToEncoding(name, false) ) ) ;
+ }
+#endif
+
+ wxMBConv_cocoa(wxFontEncoding encoding)
+ {
+ Init( wxCFStringEncFromFontEnc(encoding) );
+ }
+
+ ~wxMBConv_cocoa()
+ {
+ }
+
+ void Init( CFStringEncoding encoding)
+ {
+ m_encoding = encoding ;
+ }
+
+ size_t MB2WC(wchar_t * szOut, const char * szUnConv, size_t nOutSize) const
+ {
+ wxASSERT(szUnConv);
+
+ CFStringRef theString = CFStringCreateWithBytes (
+ NULL, //the allocator
+ (const UInt8*)szUnConv,
+ strlen(szUnConv),
+ m_encoding,
+ false //no BOM/external representation
+ );
+
+ wxASSERT(theString);
+
+ size_t nOutLength = CFStringGetLength(theString);
+
+ if (szOut == NULL)
+ {
+ CFRelease(theString);
+ return nOutLength;
+ }
+
+ CFRange theRange = { 0, nOutSize };
+
+#if SIZEOF_WCHAR_T == 4
+ UniChar* szUniCharBuffer = new UniChar[nOutSize];
+#endif
+
+ CFStringGetCharacters(theString, theRange, szUniCharBuffer);
+
+ CFRelease(theString);
+
+ szUniCharBuffer[nOutLength] = '\0' ;
+
+#if SIZEOF_WCHAR_T == 4
+ wxMBConvUTF16 converter ;
+ converter.MB2WC(szOut, (const char*)szUniCharBuffer , nOutSize ) ;
+ delete[] szUniCharBuffer;
+#endif
+
+ return nOutLength;
+ }
+
+ size_t WC2MB(char *szOut, const wchar_t *szUnConv, size_t nOutSize) const
+ {
+ wxASSERT(szUnConv);
+
+ size_t nRealOutSize;
+ size_t nBufSize = wxWcslen(szUnConv);
+ UniChar* szUniBuffer = (UniChar*) szUnConv;
+
+#if SIZEOF_WCHAR_T == 4
+ wxMBConvUTF16BE converter ;
+ nBufSize = converter.WC2MB( NULL , szUnConv , 0 );
+ szUniBuffer = new UniChar[ (nBufSize / sizeof(UniChar)) + 1] ;
+ converter.WC2MB( (char*) szUniBuffer , szUnConv, nBufSize + sizeof(UniChar)) ;
+ nBufSize /= sizeof(UniChar);
+#endif
+
+ CFStringRef theString = CFStringCreateWithCharactersNoCopy(
+ NULL, //allocator
+ szUniBuffer,
+ nBufSize,
+ kCFAllocatorNull //deallocator - we want to deallocate it ourselves
+ );
+
+ wxASSERT(theString);
+
+ //Note that CER puts a BOM when converting to unicode
+ //so we check and use getchars instead in that case
+ if (m_encoding == kCFStringEncodingUnicode)
+ {
+ if (szOut != NULL)
+ CFStringGetCharacters(theString, CFRangeMake(0, nOutSize - 1), (UniChar*) szOut);
+
+ nRealOutSize = CFStringGetLength(theString) + 1;
+ }
+ else
+ {
+ CFStringGetBytes(
+ theString,
+ CFRangeMake(0, CFStringGetLength(theString)),
+ m_encoding,
+ 0, //what to put in characters that can't be converted -
+ //0 tells CFString to return NULL if it meets such a character
+ false, //not an external representation
+ (UInt8*) szOut,
+ nOutSize,
+ (CFIndex*) &nRealOutSize
+ );
+ }
+
+ CFRelease(theString);
+
+#if SIZEOF_WCHAR_T == 4
+ delete[] szUniBuffer;
+#endif
+
+ return nRealOutSize - 1;
projects / wxWidgets.git / blobdiff