X-Git-Url: https://git.saurik.com/wxWidgets.git/blobdiff_plain/de6185e212ebc37ff11ff70278e3c4f68419b097..1a1f3e4b53fd5d1515b16905edf4250dfb2fc676:/src/common/strconv.cpp diff --git a/src/common/strconv.cpp b/src/common/strconv.cpp index eafce97c8b..843c84e046 100644 --- a/src/common/strconv.cpp +++ b/src/common/strconv.cpp @@ -15,21 +15,21 @@ // For compilers that support precompilation, includes "wx.h". #include "wx/wxprec.h" +#ifdef __BORLANDC__ + #pragma hdrstop +#endif //__BORLANDC__ + #ifndef WX_PRECOMP #include "wx/intl.h" #include "wx/log.h" #include "wx/utils.h" + #include "wx/hashmap.h" #endif #include "wx/strconv.h" #if wxUSE_WCHAR_T -#ifdef __WINDOWS__ - #include "wx/msw/private.h" - #include "wx/msw/missing.h" -#endif - #ifndef __WXWINCE__ #include #endif @@ -39,6 +39,8 @@ #include #if defined(__WIN32__) && !defined(__WXMICROWIN__) + #include "wx/msw/private.h" + #include "wx/msw/missing.h" #define wxHAVE_WIN32_MB2WC #endif @@ -422,10 +424,12 @@ wxMBConv::cMB2WC(const char *inBuff, size_t inLen, size_t *outLen) const const wxCharBuffer wxMBConv::cWC2MB(const wchar_t *inBuff, size_t inLen, size_t *outLen) const { - const size_t dstLen = FromWChar(NULL, 0, inBuff, inLen); + size_t dstLen = FromWChar(NULL, 0, inBuff, inLen); if ( dstLen != wxCONV_FAILED ) { - wxCharBuffer buf(dstLen - 1); + // special case of empty input: can't allocate 0 size buffer below as + // wxCharBuffer insists on NUL-terminating it + wxCharBuffer buf(dstLen ? dstLen - 1 : 1); if ( FromWChar(buf.data(), dstLen, inBuff, inLen) != wxCONV_FAILED ) { if ( outLen ) @@ -433,11 +437,12 @@ wxMBConv::cWC2MB(const wchar_t *inBuff, size_t inLen, size_t *outLen) const *outLen = dstLen; const size_t nulLen = GetMBNulLen(); - if ( !NotAllNULs(buf.data() + dstLen - nulLen, nulLen) ) + if ( dstLen >= nulLen && + !NotAllNULs(buf.data() + dstLen - nulLen, nulLen) ) { // in this case the output is NUL-terminated and we're not // supposed to count NUL - (*outLen) -= nulLen; + *outLen -= nulLen; } } @@ -471,11 +476,11 @@ size_t wxMBConvLibc::WC2MB(char *buf, const wchar_t *psz, size_t n) const #ifdef __UNIX__ -wxConvBrokenFileNames::wxConvBrokenFileNames(const wxChar *charset) +wxConvBrokenFileNames::wxConvBrokenFileNames(const wxString& charset) { - if ( !charset || wxStricmp(charset, _T("UTF-8")) == 0 - || wxStricmp(charset, _T("UTF8")) == 0 ) - m_conv = new wxMBConvUTF8(wxMBConvUTF8::MAP_INVALID_UTF8_TO_OCTAL); + if ( wxStricmp(charset, _T("UTF-8")) == 0 || + wxStricmp(charset, _T("UTF8")) == 0 ) + m_conv = new wxMBConvUTF8(wxMBConvUTF8::MAP_INVALID_UTF8_TO_PUA); else m_conv = new wxCSConv(charset); } @@ -1583,7 +1588,7 @@ wxMBConvUTF32swap::FromWChar(char *dst, size_t dstLen, class wxMBConv_iconv : public wxMBConv { public: - wxMBConv_iconv(const wxChar *name); + wxMBConv_iconv(const char *name); virtual ~wxMBConv_iconv(); virtual size_t MB2WC(wchar_t *buf, const char *psz, size_t n) const; @@ -1592,9 +1597,13 @@ public: // classify this encoding as explained in wxMBConv::GetMBNulLen() comment virtual size_t GetMBNulLen() const; +#if wxUSE_UNICODE_UTF8 + virtual bool IsUTF8() const; +#endif + virtual wxMBConv *Clone() const { - wxMBConv_iconv *p = new wxMBConv_iconv(m_name); + wxMBConv_iconv *p = new wxMBConv_iconv(m_name.ToAscii()); p->m_minMBCharWidth = m_minMBCharWidth; return p; } @@ -1632,7 +1641,7 @@ private: }; // make the constructor available for unit testing -WXDLLIMPEXP_BASE wxMBConv* new_wxMBConv_iconv( const wxChar* name ) +WXDLLIMPEXP_BASE wxMBConv* new_wxMBConv_iconv( const char* name ) { wxMBConv_iconv* result = new wxMBConv_iconv( name ); if ( !result->IsOk() ) @@ -1647,15 +1656,11 @@ WXDLLIMPEXP_BASE wxMBConv* new_wxMBConv_iconv( const wxChar* name ) wxString wxMBConv_iconv::ms_wcCharsetName; bool wxMBConv_iconv::ms_wcNeedsSwap = false; -wxMBConv_iconv::wxMBConv_iconv(const wxChar *name) +wxMBConv_iconv::wxMBConv_iconv(const char *name) : m_name(name) { m_minMBCharWidth = 0; - // iconv operates with chars, not wxChars, but luckily it uses only ASCII - // names for the charsets - const wxCharBuffer cname(wxString(name).ToAscii()); - // check for charset that represents wchar_t: if ( ms_wcCharsetName.empty() ) { @@ -1664,7 +1669,7 @@ wxMBConv_iconv::wxMBConv_iconv(const wxChar *name) #if wxUSE_FONTMAP const wxChar **names = wxFontMapperBase::GetAllEncodingNames(WC_ENC); #else // !wxUSE_FONTMAP - static const wxChar *names[] = + static const wxChar *names_static[] = { #if SIZEOF_WCHAR_T == 4 _T("UCS-4"), @@ -1673,6 +1678,7 @@ wxMBConv_iconv::wxMBConv_iconv(const wxChar *name) #endif NULL }; + const wxChar **names = names_static; #endif // wxUSE_FONTMAP/!wxUSE_FONTMAP for ( ; *names && ms_wcCharsetName.empty(); ++names ) @@ -1691,13 +1697,13 @@ wxMBConv_iconv::wxMBConv_iconv(const wxChar *name) wxLogTrace(TRACE_STRCONV, _T(" trying charset \"%s\""), nameXE.c_str()); - m2w = iconv_open(nameXE.ToAscii(), cname); + m2w = iconv_open(nameXE.ToAscii(), name); if ( m2w == ICONV_T_INVALID ) { // try charset w/o bytesex info (e.g. "UCS4") wxLogTrace(TRACE_STRCONV, _T(" trying charset \"%s\""), nameCS.c_str()); - m2w = iconv_open(nameCS.ToAscii(), cname); + m2w = iconv_open(nameCS.ToAscii(), name); // and check for bytesex ourselves: if ( m2w != ICONV_T_INVALID ) @@ -1740,14 +1746,14 @@ wxMBConv_iconv::wxMBConv_iconv(const wxChar *name) wxLogTrace(TRACE_STRCONV, wxT("iconv wchar_t charset is \"%s\"%s"), - ms_wcCharsetName.empty() ? _T("") - : ms_wcCharsetName.c_str(), + ms_wcCharsetName.empty() ? wxString("") + : ms_wcCharsetName, ms_wcNeedsSwap ? _T(" (needs swap)") : _T("")); } else // we already have ms_wcCharsetName { - m2w = iconv_open(ms_wcCharsetName.ToAscii(), cname); + m2w = iconv_open(ms_wcCharsetName.ToAscii(), name); } if ( ms_wcCharsetName.empty() ) @@ -1756,12 +1762,12 @@ wxMBConv_iconv::wxMBConv_iconv(const wxChar *name) } else { - w2m = iconv_open(cname, ms_wcCharsetName.ToAscii()); + w2m = iconv_open(name, ms_wcCharsetName.ToAscii()); if ( w2m == ICONV_T_INVALID ) { wxLogTrace(TRACE_STRCONV, wxT("\"%s\" -> \"%s\" works but not the converse!?"), - ms_wcCharsetName.c_str(), cname.data()); + ms_wcCharsetName.c_str(), name); } } } @@ -1802,8 +1808,8 @@ size_t wxMBConv_iconv::MB2WC(wchar_t *buf, const char *psz, size_t n) const } #if wxUSE_THREADS - // NB: iconv() is MT-safe, but each thread must use it's own iconv_t handle. - // Unfortunately there is a couple of global wxCSConv objects such as + // NB: iconv() is MT-safe, but each thread must use its own iconv_t handle. + // Unfortunately there are a couple of global wxCSConv objects such as // wxConvLocal that are used all over wx code, so we have to make sure // the handle is used by at most one thread at the time. Otherwise // only a few wx classes would be safe to use from non-main threads @@ -1950,7 +1956,7 @@ size_t wxMBConv_iconv::GetMBNulLen() const wxMutexLocker lock(self->m_iconvMutex); #endif - wchar_t *wnul = L""; + const wchar_t *wnul = L""; char buf[8]; // should be enough for NUL in any encoding size_t inLen = sizeof(wchar_t), outLen = WXSIZEOF(buf); @@ -1969,6 +1975,14 @@ size_t wxMBConv_iconv::GetMBNulLen() const return m_minMBCharWidth; } +#if wxUSE_UNICODE_UTF8 +bool wxMBConv_iconv::IsUTF8() const +{ + return wxStricmp(m_name, "UTF-8") == 0 || + wxStricmp(m_name, "UTF8") == 0; +} +#endif + #endif // HAVE_ICONV @@ -1980,7 +1994,7 @@ size_t wxMBConv_iconv::GetMBNulLen() const // from utils.cpp #if wxUSE_FONTMAP -extern WXDLLIMPEXP_BASE long wxCharsetToCodepage(const wxChar *charset); +extern WXDLLIMPEXP_BASE long wxCharsetToCodepage(const char *charset); extern WXDLLIMPEXP_BASE long wxEncodingToCodepage(wxFontEncoding encoding); #endif @@ -1994,13 +2008,14 @@ public: } wxMBConv_win32(const wxMBConv_win32& conv) + : wxMBConv() { m_CodePage = conv.m_CodePage; m_minMBCharWidth = conv.m_minMBCharWidth; } #if wxUSE_FONTMAP - wxMBConv_win32(const wxChar* name) + wxMBConv_win32(const char* name) { m_CodePage = wxCharsetToCodepage(name); m_minMBCharWidth = 0; @@ -2028,12 +2043,12 @@ public: // http://msdn.microsoft.com/library/en-us/intl/unicode_17si.asp if ( m_CodePage == CP_UTF8 ) { - return wxConvUTF8.MB2WC(buf, psz, n); + return wxMBConvUTF8().MB2WC(buf, psz, n); } if ( m_CodePage == CP_UTF7 ) { - return wxConvUTF7.MB2WC(buf, psz, n); + return wxMBConvUTF7().MB2WC(buf, psz, n); } int flags = 0; @@ -2223,11 +2238,11 @@ private: int verMaj, verMin; switch ( wxGetOsVersion(&verMaj, &verMin) ) { - case wxWIN95: + case wxOS_WINDOWS_9X: s_isWin98Or2k = verMaj >= 4 && verMin >= 10; break; - case wxWINDOWS_NT: + case wxOS_WINDOWS_NT: s_isWin98Or2k = verMaj >= 5; break; @@ -2579,7 +2594,7 @@ public: Init( wxCFStringEncFromFontEnc(encoding) ); } - ~wxMBConv_cocoa() + virtual ~wxMBConv_cocoa() { } @@ -2723,7 +2738,7 @@ public: } #if wxUSE_FONTMAP - wxMBConv_mac(const wxChar* name) + wxMBConv_mac(const char* name) { Init( wxMacGetSystemEncFromFontEnc( wxFontMapperBase::Get()->CharsetToEncoding(name, false) ) ); } @@ -2734,30 +2749,43 @@ public: Init( wxMacGetSystemEncFromFontEnc(encoding) ); } - ~wxMBConv_mac() + virtual ~wxMBConv_mac() { OSStatus status = noErr ; - status = TECDisposeConverter(m_MB2WC_converter); - status = TECDisposeConverter(m_WC2MB_converter); + if (m_MB2WC_converter) + status = TECDisposeConverter(m_MB2WC_converter); + if (m_WC2MB_converter) + status = TECDisposeConverter(m_WC2MB_converter); } - - void Init( TextEncodingBase encoding) + void Init( TextEncodingBase encoding,TextEncodingVariant encodingVariant = kTextEncodingDefaultVariant , + TextEncodingFormat encodingFormat = kTextEncodingDefaultFormat) { - OSStatus status = noErr ; - m_char_encoding = encoding ; + m_MB2WC_converter = NULL ; + m_WC2MB_converter = NULL ; + m_char_encoding = CreateTextEncoding(encoding, encodingVariant, encodingFormat) ; m_unicode_encoding = CreateTextEncoding(kTextEncodingUnicodeDefault, 0, kUnicode16BitFormat) ; + } - status = TECCreateConverter(&m_MB2WC_converter, + virtual void CreateIfNeeded() const + { + if ( m_MB2WC_converter == NULL && m_WC2MB_converter == NULL ) + { + OSStatus status = noErr ; + status = TECCreateConverter(&m_MB2WC_converter, m_char_encoding, m_unicode_encoding); - status = TECCreateConverter(&m_WC2MB_converter, + wxASSERT_MSG( status == noErr , _("Unable to create TextEncodingConverter")) ; + status = TECCreateConverter(&m_WC2MB_converter, m_unicode_encoding, m_char_encoding); + wxASSERT_MSG( status == noErr , _("Unable to create TextEncodingConverter")) ; + } } size_t MB2WC(wchar_t *buf, const char *psz, size_t n) const { + CreateIfNeeded() ; OSStatus status = noErr ; ByteCount byteOutLen ; ByteCount byteInLen = strlen(psz) + 1; @@ -2806,6 +2834,7 @@ public: size_t WC2MB(char *buf, const wchar_t *psz, size_t n) const { + CreateIfNeeded() ; OSStatus status = noErr ; ByteCount byteOutLen ; ByteCount byteInLen = wxWcslen(psz) * SIZEOF_WCHAR_T ; @@ -2868,16 +2897,192 @@ public: virtual wxMBConv *Clone() const { return new wxMBConv_mac(*this); } bool IsOk() const - { return m_MB2WC_converter != NULL && m_WC2MB_converter != NULL; } + { + CreateIfNeeded() ; + return m_MB2WC_converter != NULL && m_WC2MB_converter != NULL; + } -private: - TECObjectRef m_MB2WC_converter; - TECObjectRef m_WC2MB_converter; +protected : + mutable TECObjectRef m_MB2WC_converter; + mutable TECObjectRef m_WC2MB_converter; TextEncodingBase m_char_encoding; TextEncodingBase m_unicode_encoding; }; +// MB is decomposed (D) normalized UTF8 + +class wxMBConv_macUTF8D : public wxMBConv_mac +{ +public : + wxMBConv_macUTF8D() + { + Init( kTextEncodingUnicodeDefault , kUnicodeNoSubset , kUnicodeUTF8Format ) ; + m_uni = NULL; + m_uniBack = NULL ; + } + + virtual ~wxMBConv_macUTF8D() + { + if (m_uni!=NULL) + DisposeUnicodeToTextInfo(&m_uni); + if (m_uniBack!=NULL) + DisposeUnicodeToTextInfo(&m_uniBack); + } + + size_t WC2MB(char *buf, const wchar_t *psz, size_t n) const + { + CreateIfNeeded() ; + OSStatus status = noErr ; + ByteCount byteOutLen ; + ByteCount byteInLen = wxWcslen(psz) * SIZEOF_WCHAR_T ; + + char *tbuf = NULL ; + + if (buf == NULL) + { + // Apple specs say at least 32 + n = wxMax( 32, ((byteInLen / SIZEOF_WCHAR_T) * 8) + SIZEOF_WCHAR_T ); + tbuf = (char*) malloc( n ) ; + } + + ByteCount byteBufferLen = n ; + UniChar* ubuf = NULL ; + +#if SIZEOF_WCHAR_T == 4 + wxMBConvUTF16 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 + + // ubuf is a non-decomposed UniChar buffer + + ByteCount dcubuflen = byteInLen * 2 + 2 ; + ByteCount dcubufread , dcubufwritten ; + UniChar *dcubuf = (UniChar*) malloc( dcubuflen ) ; + + ConvertFromUnicodeToText( m_uni , byteInLen , ubuf , + kUnicodeDefaultDirectionMask, 0, NULL, NULL, NULL, dcubuflen , &dcubufread , &dcubufwritten , dcubuf ) ; + + // we now convert that decomposed buffer into UTF8 + + status = TECConvertText( + m_WC2MB_converter, (ConstTextPtr) dcubuf, dcubufwritten, &dcubufread, + (TextPtr) (buf ? buf : tbuf), byteBufferLen, &byteOutLen); + + free( dcubuf ); + +#if SIZEOF_WCHAR_T == 4 + free( ubuf ) ; +#endif + + if ( buf == NULL ) + free(tbuf) ; + + size_t res = byteOutLen ; + if ( buf && res < n) + { + buf[res] = 0; + // don't test for round-trip fidelity yet, we cannot guarantee it yet + } + + return res ; + } + + size_t MB2WC(wchar_t *buf, const char *psz, size_t n) const + { + CreateIfNeeded() ; + OSStatus status = noErr ; + ByteCount byteOutLen ; + ByteCount byteInLen = strlen(psz) + 1; + wchar_t *tbuf = NULL ; + UniChar* ubuf = NULL ; + size_t res = 0 ; + + if (buf == NULL) + { + // Apple specs say at least 32 + n = wxMax( 32, 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 + + ByteCount dcubuflen = byteBufferLen * 2 + 2 ; + ByteCount dcubufread , dcubufwritten ; + UniChar *dcubuf = (UniChar*) malloc( dcubuflen ) ; + + status = TECConvertText( + m_MB2WC_converter, (ConstTextPtr) psz, byteInLen, &byteInLen, + (TextPtr) dcubuf, dcubuflen, &byteOutLen); + // 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 + dcubuf[byteOutLen / sizeof( UniChar ) ] = 0 ; + + // now from the decomposed UniChar to properly composed uniChar + ConvertFromUnicodeToText( m_uniBack , byteOutLen , dcubuf , + kUnicodeDefaultDirectionMask, 0, NULL, NULL, NULL, dcubuflen , &dcubufread , &dcubufwritten , ubuf ) ; + + free( dcubuf ); + byteOutLen = dcubufwritten ; + ubuf[byteOutLen / sizeof( UniChar ) ] = 0 ; + + +#if SIZEOF_WCHAR_T == 4 + wxMBConvUTF16 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 ; + } + + virtual void CreateIfNeeded() const + { + wxMBConv_mac::CreateIfNeeded() ; + if ( m_uni == NULL ) + { + m_map.unicodeEncoding = CreateTextEncoding(kTextEncodingUnicodeDefault, + kUnicodeNoSubset, kTextEncodingDefaultFormat); + m_map.otherEncoding = CreateTextEncoding(kTextEncodingUnicodeDefault, + kUnicodeCanonicalDecompVariant, kTextEncodingDefaultFormat); + m_map.mappingVersion = kUnicodeUseLatestMapping; + + OSStatus err = CreateUnicodeToTextInfo(&m_map, &m_uni); + wxASSERT_MSG( err == noErr , _(" Couldn't create the UnicodeConverter")) ; + + m_map.unicodeEncoding = CreateTextEncoding(kTextEncodingUnicodeDefault, + kUnicodeNoSubset, kTextEncodingDefaultFormat); + m_map.otherEncoding = CreateTextEncoding(kTextEncodingUnicodeDefault, + kUnicodeCanonicalCompVariant, kTextEncodingDefaultFormat); + m_map.mappingVersion = kUnicodeUseLatestMapping; + err = CreateUnicodeToTextInfo(&m_map, &m_uniBack); + wxASSERT_MSG( err == noErr , _(" Couldn't create the UnicodeConverter")) ; + } + } +protected : + mutable UnicodeToTextInfo m_uni; + mutable UnicodeToTextInfo m_uniBack; + mutable UnicodeMapping m_map; +}; #endif // defined(__WXMAC__) && defined(TARGET_CARBON) // ============================================================================ @@ -2898,7 +3103,7 @@ private: public: // temporarily just use wxEncodingConverter stuff, // so that it works while a better implementation is built - wxMBConv_wxwin(const wxChar* name) + wxMBConv_wxwin(const char* name) { if (name) m_enc = wxFontMapperBase::Get()->CharsetToEncoding(name, false); @@ -2971,7 +3176,7 @@ private: }; // make the constructors available for unit testing -WXDLLIMPEXP_BASE wxMBConv* new_wxMBConv_wxwin( const wxChar* name ) +WXDLLIMPEXP_BASE wxMBConv* new_wxMBConv_wxwin( const char* name ) { wxMBConv_wxwin* result = new wxMBConv_wxwin( name ); if ( !result->IsOk() ) @@ -2996,13 +3201,13 @@ void wxCSConv::Init() m_deferred = true; } -wxCSConv::wxCSConv(const wxChar *charset) +wxCSConv::wxCSConv(const wxString& charset) { Init(); - if ( charset ) + if ( !charset.empty() ) { - SetName(charset); + SetName(charset.ToAscii()); } #if wxUSE_FONTMAP @@ -3059,17 +3264,16 @@ void wxCSConv::Clear() m_convReal = NULL; } -void wxCSConv::SetName(const wxChar *charset) +void wxCSConv::SetName(const char *charset) { if (charset) { - m_name = wxStrdup(charset); + m_name = strdup(charset); m_deferred = true; } } #if wxUSE_FONTMAP -#include "wx/hashmap.h" WX_DECLARE_HASH_MAP( wxFontEncoding, wxString, wxIntegerHash, wxIntegerEqual, wxEncodingNameCache ); @@ -3083,7 +3287,7 @@ wxMBConv *wxCSConv::DoCreate() const wxLogTrace(TRACE_STRCONV, wxT("creating conversion for %s"), (m_name ? m_name - : wxFontMapperBase::GetEncodingName(m_encoding).c_str())); + : (const char*)wxFontMapperBase::GetEncodingName(m_encoding).mb_str())); #endif // wxUSE_FONTMAP // check for the special case of ASCII or ISO8859-1 charset: as we have @@ -3110,12 +3314,13 @@ wxMBConv *wxCSConv::DoCreate() const if ( m_name ) #endif // !wxUSE_FONTMAP { - wxString name(m_name); +#if wxUSE_FONTMAP wxFontEncoding encoding(m_encoding); +#endif - if ( !name.empty() ) + if ( m_name ) { - wxMBConv_iconv *conv = new wxMBConv_iconv(name); + wxMBConv_iconv *conv = new wxMBConv_iconv(m_name); if ( conv->IsOk() ) return conv; @@ -3123,7 +3328,7 @@ wxMBConv *wxCSConv::DoCreate() const #if wxUSE_FONTMAP encoding = - wxFontMapperBase::Get()->CharsetToEncoding(name, false); + wxFontMapperBase::Get()->CharsetToEncoding(m_name, false); #endif // wxUSE_FONTMAP } #if wxUSE_FONTMAP @@ -3134,7 +3339,7 @@ wxMBConv *wxCSConv::DoCreate() const if ( it->second.empty() ) return NULL; - wxMBConv_iconv *conv = new wxMBConv_iconv(it->second); + wxMBConv_iconv *conv = new wxMBConv_iconv(it->second.ToAscii()); if ( conv->IsOk() ) return conv; @@ -3142,20 +3347,30 @@ wxMBConv *wxCSConv::DoCreate() const } const wxChar** names = wxFontMapperBase::GetAllEncodingNames(encoding); - - for ( ; *names; ++names ) + // CS : in case this does not return valid names (eg for MacRoman) + // encoding got a 'failure' entry in the cache all the same, + // although it just has to be created using a different method, so + // only store failed iconv creation attempts (or perhaps we + // shoulnd't do this at all ?) + if ( names[0] != NULL ) { - wxMBConv_iconv *conv = new wxMBConv_iconv(*names); - if ( conv->IsOk() ) + for ( ; *names; ++names ) { - gs_nameCache[encoding] = *names; - return conv; + // FIXME-UTF8: wxFontMapperBase::GetAllEncodingNames() + // will need changes that will obsolete this + wxString name(*names); + wxMBConv_iconv *conv = new wxMBConv_iconv(name.ToAscii()); + if ( conv->IsOk() ) + { + gs_nameCache[encoding] = *names; + return conv; + } + + delete conv; } - delete conv; + gs_nameCache[encoding] = _T(""); // cache the failure } - - gs_nameCache[encoding] = _T(""); // cache the failure } #endif // wxUSE_FONTMAP } @@ -3266,10 +3481,10 @@ wxMBConv *wxCSConv::DoCreate() const // 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. + // attach a timestamp, for which it will need wxConvLocal (to convert + // time to char* and then wchar_t*), but that fails, tries to log the + // error, but wxLog has an (already locked) critical section that + // guards the static buffer. static bool alreadyLoggingError = false; if (!alreadyLoggingError) { @@ -3278,9 +3493,9 @@ wxMBConv *wxCSConv::DoCreate() const m_name ? m_name : #if wxUSE_FONTMAP - wxFontMapperBase::GetEncodingDescription(m_encoding).c_str() + (const char*)wxFontMapperBase::GetEncodingDescription(m_encoding).ToAscii() #else // !wxUSE_FONTMAP - wxString::Format(_("encoding %s"), m_encoding).c_str() + (const char*)wxString::Format(_("encoding %i"), m_encoding).ToAscii() #endif // wxUSE_FONTMAP/!wxUSE_FONTMAP ); @@ -3296,20 +3511,60 @@ void wxCSConv::CreateConvIfNeeded() const { 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 ) { - self->m_name = wxStrdup(wxLocale::GetSystemEncodingName()); - } +#if wxUSE_INTL + self->m_encoding = wxLocale::GetSystemEncoding(); +#else + // fallback to some reasonable default: + self->m_encoding = wxFONTENCODING_ISO8859_1; #endif // wxUSE_INTL + } self->m_convReal = DoCreate(); self->m_deferred = false; } } +bool wxCSConv::IsOk() const +{ + CreateConvIfNeeded(); + + // special case: no convReal created for wxFONTENCODING_ISO8859_1 + if ( m_encoding == wxFONTENCODING_ISO8859_1 ) + return true; // always ok as we do it ourselves + + // m_convReal->IsOk() is called at its own creation, so we know it must + // be ok if m_convReal is non-NULL + return m_convReal != NULL; +} + +size_t wxCSConv::ToWChar(wchar_t *dst, size_t dstLen, + const char *src, size_t srcLen) const +{ + CreateConvIfNeeded(); + + if (m_convReal) + return m_convReal->ToWChar(dst, dstLen, src, srcLen); + + // latin-1 (direct) + return wxMBConv::ToWChar(dst, dstLen, src, srcLen); +} + +size_t wxCSConv::FromWChar(char *dst, size_t dstLen, + const wchar_t *src, size_t srcLen) const +{ + CreateConvIfNeeded(); + + if (m_convReal) + return m_convReal->FromWChar(dst, dstLen, src, srcLen); + + // latin-1 (direct) + return wxMBConv::FromWChar(dst, dstLen, src, srcLen); +} + size_t wxCSConv::MB2WC(wchar_t *buf, const char *psz, size_t n) const { CreateConvIfNeeded(); @@ -3369,42 +3624,122 @@ size_t wxCSConv::GetMBNulLen() const return m_convReal->GetMBNulLen(); } + // otherwise, we are ISO-8859-1 return 1; } +#if wxUSE_UNICODE_UTF8 +bool wxCSConv::IsUTF8() const +{ + CreateConvIfNeeded(); + + if ( m_convReal ) + { + return m_convReal->IsUTF8(); + } + + // otherwise, we are ISO-8859-1 + return false; +} +#endif + + +#if wxUSE_UNICODE + +wxWCharBuffer wxSafeConvertMB2WX(const char *s) +{ + if ( !s ) + return wxWCharBuffer(); + + wxWCharBuffer wbuf(wxConvLibc.cMB2WX(s)); + if ( !wbuf ) + wbuf = wxMBConvUTF8().cMB2WX(s); + if ( !wbuf ) + wbuf = wxConvISO8859_1.cMB2WX(s); + + return wbuf; +} + +wxCharBuffer wxSafeConvertWX2MB(const wchar_t *ws) +{ + if ( !ws ) + return wxCharBuffer(); + + wxCharBuffer buf(wxConvLibc.cWX2MB(ws)); + if ( !buf ) + buf = wxMBConvUTF8(wxMBConvUTF8::MAP_INVALID_UTF8_TO_OCTAL).cWX2MB(ws); + + return buf; +} + +#endif // wxUSE_UNICODE + // ---------------------------------------------------------------------------- // globals // ---------------------------------------------------------------------------- +// NB: The reason why we create converted objects in this convoluted way, +// using a factory function instead of global variable, is that they +// may be used at static initialization time (some of them are used by +// wxString ctors and there may be a global wxString object). In other +// words, possibly _before_ the converter global object would be +// initialized. + +#undef wxConvLibc +#undef wxConvUTF8 +#undef wxConvUTF7 +#undef wxConvLocal +#undef wxConvISO8859_1 + +#define WX_DEFINE_GLOBAL_CONV2(klass, impl_klass, name, ctor_args) \ + WXDLLIMPEXP_DATA_BASE(klass*) name##Ptr = NULL; \ + WXDLLIMPEXP_BASE klass* wxGet_##name##Ptr() \ + { \ + static impl_klass name##Obj ctor_args; \ + return &name##Obj; \ + } \ + /* this ensures that all global converter objects are created */ \ + /* by the time static initialization is done, i.e. before any */ \ + /* thread is launched: */ \ + static klass* gs_##name##instance = wxGet_##name##Ptr() + +#define WX_DEFINE_GLOBAL_CONV(klass, name, ctor_args) \ + WX_DEFINE_GLOBAL_CONV2(klass, klass, name, ctor_args) + #ifdef __WINDOWS__ - static wxMBConv_win32 wxConvLibcObj; + WX_DEFINE_GLOBAL_CONV2(wxMBConv, wxMBConv_win32, wxConvLibc, wxEMPTY_PARAMETER_VALUE); #elif defined(__WXMAC__) && !defined(__MACH__) - static wxMBConv_mac wxConvLibcObj ; + WX_DEFINE_GLOBAL_CONV2(wxMBConv, wxMBConv_mac, wxConvLibc, wxEMPTY_PARAMETER_VALUE); #else - static wxMBConvLibc wxConvLibcObj; + WX_DEFINE_GLOBAL_CONV2(wxMBConv, wxMBConvLibc, wxConvLibc, wxEMPTY_PARAMETER_VALUE); #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; -WXDLLIMPEXP_DATA_BASE(wxMBConv *) wxConvUI = &wxConvLocal; -WXDLLIMPEXP_DATA_BASE(wxMBConv *) wxConvFileName = & +WX_DEFINE_GLOBAL_CONV(wxMBConvUTF8, wxConvUTF8, wxEMPTY_PARAMETER_VALUE); +WX_DEFINE_GLOBAL_CONV(wxMBConvUTF7, wxConvUTF7, wxEMPTY_PARAMETER_VALUE); + +WX_DEFINE_GLOBAL_CONV(wxCSConv, wxConvLocal, (wxFONTENCODING_SYSTEM)); +WX_DEFINE_GLOBAL_CONV(wxCSConv, wxConvISO8859_1, (wxFONTENCODING_ISO8859_1)); + +WXDLLIMPEXP_DATA_BASE(wxMBConv *) wxConvCurrent = wxGet_wxConvLibcPtr(); +WXDLLIMPEXP_DATA_BASE(wxMBConv *) wxConvUI = wxGet_wxConvLocalPtr(); + +#if defined(__WXMAC__) && defined(TARGET_CARBON) +static wxMBConv_macUTF8D wxConvMacUTF8DObj; +#endif +WXDLLIMPEXP_DATA_BASE(wxMBConv *) wxConvFileName = #ifdef __WXOSX__ - wxConvUTF8Obj; +#if defined(__WXMAC__) && defined(TARGET_CARBON) + &wxConvMacUTF8DObj; #else - wxConvLibcObj; + wxGet_wxConvUTF8Ptr(); #endif +#else // !__WXOSX__ + wxGet_wxConvLibcPtr(); +#endif // __WXOSX__/!__WXOSX__ #else // !wxUSE_WCHAR_T +// FIXME-UTF8: remove this, wxUSE_WCHAR_T is required now // stand-ins in absence of wchar_t WXDLLIMPEXP_DATA_BASE(wxMBConv) wxConvLibc, wxConvISO8859_1,