X-Git-Url: https://git.saurik.com/wxWidgets.git/blobdiff_plain/830f8f11bca5c0892ae767ba14790c8b5b59011f..82df62ef8e55adf38221afc82439edeed4fe3368:/src/common/strconv.cpp diff --git a/src/common/strconv.cpp b/src/common/strconv.cpp index 3670582926..95ef424042 100644 --- a/src/common/strconv.cpp +++ b/src/common/strconv.cpp @@ -12,25 +12,18 @@ // Licence: wxWindows licence ///////////////////////////////////////////////////////////////////////////// -// ============================================================================ -// declarations -// ============================================================================ - -// ---------------------------------------------------------------------------- -// headers -// ---------------------------------------------------------------------------- - // For compilers that support precompilation, includes "wx.h". #include "wx/wxprec.h" -#ifdef __BORLANDC__ - #pragma hdrstop -#endif - #ifndef WX_PRECOMP + #ifdef __WXMSW__ + #include "wx/msw/missing.h" + #endif #include "wx/intl.h" #include "wx/log.h" -#endif // WX_PRECOMP + #include "wx/utils.h" + #include "wx/hashmap.h" +#endif #include "wx/strconv.h" @@ -38,7 +31,6 @@ #ifdef __WINDOWS__ #include "wx/msw/private.h" - #include "wx/msw/missing.h" #endif #ifndef __WXWINCE__ @@ -51,7 +43,7 @@ #if defined(__WIN32__) && !defined(__WXMICROWIN__) #define wxHAVE_WIN32_MB2WC -#endif // __WIN32__ but !__WXMICROWIN__ +#endif #ifdef __SALFORDC__ #include @@ -64,7 +56,6 @@ #include "wx/encconv.h" #include "wx/fontmap.h" -#include "wx/utils.h" #ifdef __WXMAC__ #ifndef __DARWIN__ @@ -73,15 +64,20 @@ #include #endif -#include "wx/mac/private.h" // includes mac headers +// includes Mac headers +#include "wx/mac/private.h" #endif + #define TRACE_STRCONV _T("strconv") +// WC_UTF16 is defined only if sizeof(wchar_t) == 2, otherwise it's supposed to +// be 4 bytes #if SIZEOF_WCHAR_T == 2 #define WC_UTF16 #endif + // ============================================================================ // implementation // ============================================================================ @@ -96,44 +92,45 @@ static bool NotAllNULs(const char *p, size_t n) } // ---------------------------------------------------------------------------- -// UTF-16 en/decoding to/from UCS-4 +// UTF-16 en/decoding to/from UCS-4 with surrogates handling // ---------------------------------------------------------------------------- - static size_t encode_utf16(wxUint32 input, wxUint16 *output) { - if (input<=0xffff) + if (input <= 0xffff) { if (output) *output = (wxUint16) input; + return 1; } - else if (input>=0x110000) + else if (input >= 0x110000) { - return (size_t)-1; + return wxCONV_FAILED; } else { if (output) { - *output++ = (wxUint16) ((input >> 10)+0xd7c0); - *output = (wxUint16) ((input&0x3ff)+0xdc00); + *output++ = (wxUint16) ((input >> 10) + 0xd7c0); + *output = (wxUint16) ((input & 0x3ff) + 0xdc00); } + return 2; } } static size_t decode_utf16(const wxUint16* input, wxUint32& output) { - if ((*input<0xd800) || (*input>0xdfff)) + if ((*input < 0xd800) || (*input > 0xdfff)) { output = *input; return 1; } - else if ((input[1]<0xdc00) || (input[1]>0xdfff)) + else if ((input[1] < 0xdc00) || (input[1] > 0xdfff)) { output = *input; - return (size_t)-1; + return wxCONV_FAILED; } else { @@ -142,6 +139,29 @@ static size_t decode_utf16(const wxUint16* input, wxUint32& output) } } +#ifdef WC_UTF16 + typedef wchar_t wxDecodeSurrogate_t; +#else // !WC_UTF16 + typedef wxUint16 wxDecodeSurrogate_t; +#endif // WC_UTF16/!WC_UTF16 + +// returns the next UTF-32 character from the wchar_t buffer and advances the +// pointer to the character after this one +// +// if an invalid character is found, *pSrc is set to NULL, the caller must +// check for this +static wxUint32 wxDecodeSurrogate(const wxDecodeSurrogate_t **pSrc) +{ + wxUint32 out; + const size_t + n = decode_utf16(wx_reinterpret_cast(const wxUint16 *, *pSrc), out); + if ( n == wxCONV_FAILED ) + *pSrc = NULL; + else + *pSrc += n; + + return out; +} // ---------------------------------------------------------------------------- // wxMBConv @@ -162,7 +182,7 @@ wxMBConv::ToWChar(wchar_t *dst, size_t dstLen, size_t dstWritten = 0; // the number of NULs terminating this string - size_t nulLen wxDUMMY_INITIALIZE(0); + size_t nulLen = 0; // not really needed, but just to avoid warnings // if we were not given the input size we just have to assume that the // string is properly terminated as we have no way of knowing how long it @@ -170,7 +190,7 @@ wxMBConv::ToWChar(wchar_t *dst, size_t dstLen, // NULs at the end wxCharBuffer bufTmp; const char *srcEnd; - if ( srcLen != (size_t)-1 ) + if ( srcLen != wxNO_LEN ) { // we need to know how to find the end of this string nulLen = GetMBNulLen(); @@ -201,21 +221,19 @@ wxMBConv::ToWChar(wchar_t *dst, size_t dstLen, { // try to convert the current chunk size_t lenChunk = MB2WC(NULL, src, 0); - if ( lenChunk == 0 ) - { - // nothing left in the input string, conversion succeeded; but - // still account for the trailing NULL - dstWritten++; - break; - } - if ( lenChunk == wxCONV_FAILED ) return wxCONV_FAILED; - lenChunk++; // for trailing NUL + lenChunk++; // for the L'\0' at the end of this chunk dstWritten += lenChunk; + if ( lenChunk == 1 ) + { + // nothing left in the input string, conversion succeeded + break; + } + if ( dst ) { if ( dstWritten > dstLen ) @@ -229,8 +247,8 @@ wxMBConv::ToWChar(wchar_t *dst, size_t dstLen, if ( !srcEnd ) { - // we convert the entire string in this cas, as we suppose that the - // string is NUL-terminated and so srcEnd is not used at all + // we convert just one chunk in this case as this is the entire + // string anyhow break; } @@ -269,7 +287,7 @@ wxMBConv::FromWChar(char *dst, size_t dstLen, // if we don't know its length we have no choice but to assume that it is, // indeed, properly terminated wxWCharBuffer bufTmp; - if ( srcLen == (size_t)-1 ) + if ( srcLen == wxNO_LEN ) { srcLen = wxWcslen(src) + 1; } @@ -277,7 +295,7 @@ wxMBConv::FromWChar(char *dst, size_t dstLen, { // make a copy in order to properly NUL-terminate the string bufTmp = wxWCharBuffer(srcLen); - memcpy(bufTmp.data(), src, srcLen*sizeof(wchar_t)); + memcpy(bufTmp.data(), src, srcLen * sizeof(wchar_t)); src = bufTmp; } @@ -310,9 +328,9 @@ wxMBConv::FromWChar(char *dst, size_t dstLen, return dstWritten; } -size_t wxMBConv::MB2WC(wchar_t *out, const char *in, size_t outLen) const +size_t wxMBConv::MB2WC(wchar_t *outBuff, const char *inBuff, size_t outLen) const { - size_t rc = ToWChar(out, outLen, in); + size_t rc = ToWChar(outBuff, outLen, inBuff); if ( rc != wxCONV_FAILED ) { // ToWChar() returns the buffer length, i.e. including the trailing @@ -323,9 +341,9 @@ size_t wxMBConv::MB2WC(wchar_t *out, const char *in, size_t outLen) const return rc; } -size_t wxMBConv::WC2MB(char *out, const wchar_t *in, size_t outLen) const +size_t wxMBConv::WC2MB(char *outBuff, const wchar_t *inBuff, size_t outLen) const { - size_t rc = FromWChar(out, outLen, in); + size_t rc = FromWChar(outBuff, outLen, inBuff); if ( rc != wxCONV_FAILED ) { rc -= GetMBNulLen(); @@ -379,16 +397,21 @@ const wxCharBuffer wxMBConv::cWC2MB(const wchar_t *pwz) const } const wxWCharBuffer -wxMBConv::cMB2WC(const char *in, size_t inLen, size_t *outLen) const +wxMBConv::cMB2WC(const char *inBuff, size_t inLen, size_t *outLen) const { - const size_t dstLen = ToWChar(NULL, 0, in, inLen); + const size_t dstLen = ToWChar(NULL, 0, inBuff, inLen); if ( dstLen != wxCONV_FAILED ) { wxWCharBuffer wbuf(dstLen - 1); - if ( ToWChar(wbuf.data(), dstLen, in, inLen) ) + if ( ToWChar(wbuf.data(), dstLen, inBuff, inLen) != wxCONV_FAILED ) { if ( outLen ) - *outLen = dstLen - 1; + { + *outLen = dstLen; + if ( wbuf[dstLen - 1] == L'\0' ) + (*outLen)--; + } + return wbuf; } } @@ -400,16 +423,30 @@ wxMBConv::cMB2WC(const char *in, size_t inLen, size_t *outLen) const } const wxCharBuffer -wxMBConv::cWC2MB(const wchar_t *in, size_t inLen, size_t *outLen) const +wxMBConv::cWC2MB(const wchar_t *inBuff, size_t inLen, size_t *outLen) const { - const size_t dstLen = FromWChar(NULL, 0, in, inLen); + size_t dstLen = FromWChar(NULL, 0, inBuff, inLen); if ( dstLen != wxCONV_FAILED ) { - wxCharBuffer buf(dstLen - 1); - if ( FromWChar(buf.data(), dstLen, in, inLen) ) + // 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 ) - *outLen = dstLen - 1; + { + *outLen = dstLen; + + const size_t nulLen = GetMBNulLen(); + 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; + } + } + return buf; } } @@ -550,7 +587,7 @@ size_t wxMBConvUTF7::MB2WC(wchar_t *buf, const char *psz, size_t n) const if ( !ok ) { // in valid UTF7 we should have valid characters after '+' - return (size_t)-1; + return wxCONV_FAILED; } if (*psz == '-') @@ -611,19 +648,21 @@ size_t wxMBConvUTF7::WC2MB(char *buf, const wchar_t *psz, size_t n) const // plain ASCII char if (buf) *buf++ = (char)cc; + len++; } #ifndef WC_UTF16 else if (((wxUint32)cc) > 0xffff) { // no surrogate pair generation (yet?) - return (size_t)-1; + return wxCONV_FAILED; } #endif else { if (buf) *buf++ = '+'; + len++; if (cc != '+') { @@ -644,24 +683,30 @@ size_t wxMBConvUTF7::WC2MB(char *buf, const wchar_t *psz, size_t n) const len++; } } + cc = *psz; if (!(cc) || (cc < 0x80 && utf7encode[cc] < 1)) break; } + if (l != 0) { if (buf) *buf++ = utf7enb64[((d % 16) << (6 - l)) % 64]; + len++; } } + if (buf) *buf++ = '-'; len++; } } + if (buf && (len < n)) *buf = 0; + return len; } @@ -689,6 +734,7 @@ size_t wxMBConvUTF8::MB2WC(wchar_t *buf, const char *psz, size_t n) const unsigned cnt; for (cnt = 0; fc & 0x80; cnt++) fc <<= 1; + if (!cnt) { // plain ASCII char @@ -726,9 +772,11 @@ size_t wxMBConvUTF8::MB2WC(wchar_t *buf, const char *psz, size_t n) const invalid = true; break; } + psz++; res = (res << 6) | (cc & 0x3f); } + if (invalid || res <= utf8_max[ocnt]) { // illegal UTF-8 encoding @@ -747,7 +795,7 @@ size_t wxMBConvUTF8::MB2WC(wchar_t *buf, const char *psz, size_t n) const #ifdef WC_UTF16 // cast is ok because wchar_t == wxUuint16 if WC_UTF16 size_t pa = encode_utf16(res, (wxUint16 *)buf); - if (pa == (size_t)-1) + if (pa == wxCONV_FAILED) { invalid = true; } @@ -764,6 +812,7 @@ size_t wxMBConvUTF8::MB2WC(wchar_t *buf, const char *psz, size_t n) const #endif // WC_UTF16/!WC_UTF16 } } + if (invalid) { if (m_options & MAP_INVALID_UTF8_TO_PUA) @@ -773,7 +822,7 @@ size_t wxMBConvUTF8::MB2WC(wchar_t *buf, const char *psz, size_t n) const #ifdef WC_UTF16 // cast is ok because wchar_t == wxUuint16 if WC_UTF16 size_t pa = encode_utf16((unsigned char)*opsz + wxUnicodePUA, (wxUint16 *)buf); - wxASSERT(pa != (size_t)-1); + wxASSERT(pa != wxCONV_FAILED); if (buf) buf += pa; opsz++; @@ -798,19 +847,22 @@ size_t wxMBConvUTF8::MB2WC(wchar_t *buf, const char *psz, size_t n) const *buf++ = (wchar_t)( L'0' + (on % 0100) / 010 ); *buf++ = (wchar_t)( L'0' + on % 010 ); } + opsz++; len += 4; } } else // MAP_INVALID_UTF8_NOT { - return (size_t)-1; + return wxCONV_FAILED; } } } } + if (buf && (len < n)) *buf = 0; + return len; } @@ -826,12 +878,13 @@ size_t wxMBConvUTF8::WC2MB(char *buf, const wchar_t *psz, size_t n) const while (*psz && ((!buf) || (len < n))) { wxUint32 cc; + #ifdef WC_UTF16 // cast is ok for WC_UTF16 size_t pa = decode_utf16((const wxUint16 *)psz, cc); - psz += (pa == (size_t)-1) ? 1 : pa; + psz += (pa == wxCONV_FAILED) ? 1 : pa; #else - cc=(*psz++) & 0x7fffffff; + cc = (*psz++) & 0x7fffffff; #endif if ( (m_options & MAP_INVALID_UTF8_TO_PUA) @@ -855,8 +908,8 @@ size_t wxMBConvUTF8::WC2MB(char *buf, const wchar_t *psz, size_t n) const { if (buf) { - *buf++ = (char) ((psz[0] - L'0')*0100 + - (psz[1] - L'0')*010 + + *buf++ = (char) ((psz[0] - L'0') * 0100 + + (psz[1] - L'0') * 010 + (psz[2] - L'0')); } @@ -866,7 +919,10 @@ size_t wxMBConvUTF8::WC2MB(char *buf, const wchar_t *psz, size_t n) const else { unsigned cnt; - for (cnt = 0; cc > utf8_max[cnt]; cnt++) {} + for (cnt = 0; cc > utf8_max[cnt]; cnt++) + { + } + if (!cnt) { // plain ASCII char @@ -874,7 +930,6 @@ size_t wxMBConvUTF8::WC2MB(char *buf, const wchar_t *psz, size_t n) const *buf++ = (char) cc; len++; } - else { len += cnt + 1; @@ -888,15 +943,15 @@ size_t wxMBConvUTF8::WC2MB(char *buf, const wchar_t *psz, size_t n) const } } - if (buf && (len dstLen ) + return wxCONV_FAILED; + + *dst++ = ch; } - if (buf && len 1) + if ( outLen > dstLen ) + return wxCONV_FAILED; + + *outBuff++ = cc[0]; + if ( numChars == 2 ) { - *(wxUint16*)buf = cc[1]; - buf += sizeof(wxUint16); + // second character of a surrogate + *outBuff++ = cc[1]; } } - - len += pa*sizeof(wxUint16); - psz++; } - if (buf && len<=n-sizeof(wxUint16)) *(wxUint16*)buf=0; - return len; + return outLen; } +// ---------------------------------------------------------------------------- +// endian-reversing conversions +// ---------------------------------------------------------------------------- -// swap 16bit MB to 32bit String -size_t wxMBConvUTF16swap::MB2WC(wchar_t *buf, const char *psz, size_t n) const +size_t +wxMBConvUTF16swap::ToWChar(wchar_t *dst, size_t dstLen, + const char *src, size_t srcLen) const { - size_t len=0; + srcLen = GetLength(src, srcLen); + if ( srcLen == wxNO_LEN ) + return wxCONV_FAILED; - while (*(wxUint16*)psz && (!buf || len < n)) + const size_t inLen = srcLen / BYTES_PER_CHAR; + if ( !dst ) { - wxUint32 cc; - char tmp[4]; - tmp[0]=psz[1]; tmp[1]=psz[0]; - tmp[2]=psz[3]; tmp[3]=psz[2]; + // optimization: return maximal space which could be needed for this + // string even if the real size could be smaller if the buffer contains + // any surrogates + return inLen; + } - size_t pa=decode_utf16((wxUint16*)tmp, cc); - if (pa == (size_t)-1) - return pa; + size_t outLen = 0; + const wxUint16 *inBuff = wx_reinterpret_cast(const wxUint16 *, src); + for ( const wxUint16 * const inEnd = inBuff + inLen; inBuff < inEnd; ) + { + wxUint32 ch; + wxUint16 tmp[2]; - if (buf) - *buf++ = (wchar_t)cc; + tmp[0] = wxUINT16_SWAP_ALWAYS(*inBuff); + inBuff++; + tmp[1] = wxUINT16_SWAP_ALWAYS(*inBuff); - len++; - psz += pa * sizeof(wxUint16); + const size_t numChars = decode_utf16(tmp, ch); + if ( numChars == wxCONV_FAILED ) + return wxCONV_FAILED; + + if ( numChars == 2 ) + inBuff++; + + if ( ++outLen > dstLen ) + return wxCONV_FAILED; + + *dst++ = ch; } - if (buf && len 1) + if ( outLen > dstLen ) + return wxCONV_FAILED; + + *outBuff++ = wxUINT16_SWAP_ALWAYS(cc[0]); + if ( numChars == 2 ) { - *buf++ = ((char*)cc)[3]; - *buf++ = ((char*)cc)[2]; + // second character of a surrogate + *outBuff++ = wxUINT16_SWAP_ALWAYS(cc[1]); } } - - len += pa*sizeof(wxUint16); - psz++; } - if (buf && len<=n-sizeof(wxUint16)) *(wxUint16*)buf=0; - return len; + return outLen; } -#endif // WC_UTF16 +#endif // WC_UTF16/!WC_UTF16 -// ---------------------------------------------------------------------------- +// ============================================================================ // UTF-32 -// ---------------------------------------------------------------------------- +// ============================================================================ #ifdef WORDS_BIGENDIAN -#define wxMBConvUTF32straight wxMBConvUTF32BE -#define wxMBConvUTF32swap wxMBConvUTF32LE + #define wxMBConvUTF32straight wxMBConvUTF32BE + #define wxMBConvUTF32swap wxMBConvUTF32LE #else -#define wxMBConvUTF32swap wxMBConvUTF32BE -#define wxMBConvUTF32straight wxMBConvUTF32LE + #define wxMBConvUTF32swap wxMBConvUTF32BE + #define wxMBConvUTF32straight wxMBConvUTF32LE #endif WXDLLIMPEXP_DATA_BASE(wxMBConvUTF32LE) wxConvUTF32LE; WXDLLIMPEXP_DATA_BASE(wxMBConvUTF32BE) wxConvUTF32BE; +/* static */ +size_t wxMBConvUTF32Base::GetLength(const char *src, size_t srcLen) +{ + if ( srcLen == wxNO_LEN ) + { + // count the number of bytes in input, including the trailing NULs + const wxUint32 *inBuff = wx_reinterpret_cast(const wxUint32 *, src); + for ( srcLen = 1; *inBuff++; srcLen++ ) + ; + + srcLen *= BYTES_PER_CHAR; + } + else // we already have the length + { + // we can only convert an entire number of UTF-32 characters + if ( srcLen % BYTES_PER_CHAR ) + return wxCONV_FAILED; + } + return srcLen; +} + +// case when in-memory representation is UTF-16 #ifdef WC_UTF16 -// copy 32bit MB to 16bit String -size_t wxMBConvUTF32straight::MB2WC(wchar_t *buf, const char *psz, size_t n) const +// ---------------------------------------------------------------------------- +// conversions without endianness change +// ---------------------------------------------------------------------------- + +size_t +wxMBConvUTF32straight::ToWChar(wchar_t *dst, size_t dstLen, + const char *src, size_t srcLen) const { - size_t len=0; + srcLen = GetLength(src, srcLen); + if ( srcLen == wxNO_LEN ) + return wxCONV_FAILED; - while (*(wxUint32*)psz && (!buf || len < n)) + const wxUint32 *inBuff = wx_reinterpret_cast(const wxUint32 *, src); + const size_t inLen = srcLen / BYTES_PER_CHAR; + size_t outLen = 0; + for ( size_t n = 0; n < inLen; n++ ) { wxUint16 cc[2]; + const size_t numChars = encode_utf16(*inBuff++, cc); + if ( numChars == wxCONV_FAILED ) + return wxCONV_FAILED; - size_t pa=encode_utf16(*(wxUint32*)psz, cc); - if (pa == (size_t)-1) - return pa; - - if (buf) + outLen += numChars; + if ( dst ) { - *buf++ = cc[0]; - if (pa > 1) - *buf++ = cc[1]; + if ( outLen > dstLen ) + return wxCONV_FAILED; + + *dst++ = cc[0]; + if ( numChars == 2 ) + { + // second character of a surrogate + *dst++ = cc[1]; + } } - len += pa; - psz += sizeof(wxUint32); } - if (buf && len dstLen ) + return wxCONV_FAILED; - return len; -} + *outBuff++ = ch; + } + return outLen; +} +// ---------------------------------------------------------------------------- +// endian-reversing conversions +// ---------------------------------------------------------------------------- -// swap 32bit MB to 16bit String -size_t wxMBConvUTF32swap::MB2WC(wchar_t *buf, const char *psz, size_t n) const +size_t +wxMBConvUTF32swap::ToWChar(wchar_t *dst, size_t dstLen, + const char *src, size_t srcLen) const { - size_t len=0; + srcLen = GetLength(src, srcLen); + if ( srcLen == wxNO_LEN ) + return wxCONV_FAILED; - while (*(wxUint32*)psz && (!buf || len < n)) + const wxUint32 *inBuff = wx_reinterpret_cast(const wxUint32 *, src); + const size_t inLen = srcLen / BYTES_PER_CHAR; + size_t outLen = 0; + for ( size_t n = 0; n < inLen; n++, inBuff++ ) { - char tmp[4]; - tmp[0] = psz[3]; tmp[1] = psz[2]; - tmp[2] = psz[1]; tmp[3] = psz[0]; - - wxUint16 cc[2]; + const size_t numChars = encode_utf16(wxUINT32_SWAP_ALWAYS(*inBuff), cc); + if ( numChars == wxCONV_FAILED ) + return wxCONV_FAILED; - size_t pa=encode_utf16(*(wxUint32*)tmp, cc); - if (pa == (size_t)-1) - return pa; - - if (buf) + outLen += numChars; + if ( dst ) { - *buf++ = cc[0]; - if (pa > 1) - *buf++ = cc[1]; + if ( outLen > dstLen ) + return wxCONV_FAILED; + + *dst++ = cc[0]; + if ( numChars == 2 ) + { + // second character of a surrogate + *dst++ = cc[1]; + } } - len += pa; - psz += sizeof(wxUint32); } - if (buf && len dstLen ) + return wxCONV_FAILED; - if (buf && len<=n-sizeof(wxUint32)) - *(wxUint32*)buf=0; + *outBuff++ = wxUINT32_SWAP_ALWAYS(ch); + } - return len; + return outLen; } -#else // WC_UTF16 +#else // !WC_UTF16: wchar_t is UTF-32 +// ---------------------------------------------------------------------------- +// conversions without endianness change +// ---------------------------------------------------------------------------- -// copy 32bit MB to 32bit String -size_t wxMBConvUTF32straight::MB2WC(wchar_t *buf, const char *psz, size_t n) const +size_t +wxMBConvUTF32straight::ToWChar(wchar_t *dst, size_t dstLen, + const char *src, size_t srcLen) const { - size_t len=0; + // use memcpy() as it should be much faster than hand-written loop + srcLen = GetLength(src, srcLen); + if ( srcLen == wxNO_LEN ) + return wxCONV_FAILED; - while (*(wxUint32*)psz && (!buf || len < n)) + const size_t inLen = srcLen/BYTES_PER_CHAR; + if ( dst ) { - if (buf) - *buf++ = (wchar_t)(*(wxUint32*)psz); - len++; - psz += sizeof(wxUint32); - } + if ( dstLen < inLen ) + return wxCONV_FAILED; - if (buf && lenm_minMBCharWidth = m_minMBCharWidth; + return p; + } + bool IsOk() const { return (m2w != ICONV_T_INVALID) && (w2m != ICONV_T_INVALID); } protected: - // the iconv handlers used to translate from multibyte to wide char and in - // the other direction + // the iconv handlers used to translate from multibyte + // to wide char and in the other direction iconv_t m2w, w2m; + #if wxUSE_THREADS // guards access to m2w and w2m objects wxMutex m_iconvMutex; @@ -1440,6 +1628,10 @@ private: // different endian-ness than the native one static bool ms_wcNeedsSwap; + + // name of the encoding handled by this conversion + wxString m_name; + // cached result of GetMBNulLen(); set to 0 meaning "unknown" // initially size_t m_minMBCharWidth; @@ -1454,6 +1646,7 @@ WXDLLIMPEXP_BASE wxMBConv* new_wxMBConv_iconv( const wxChar* name ) delete result; return 0; } + return result; } @@ -1461,6 +1654,7 @@ wxString wxMBConv_iconv::ms_wcCharsetName; bool wxMBConv_iconv::ms_wcNeedsSwap = false; wxMBConv_iconv::wxMBConv_iconv(const wxChar *name) + : m_name(name) { m_minMBCharWidth = 0; @@ -1476,7 +1670,7 @@ 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"), @@ -1485,6 +1679,7 @@ 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 ) @@ -1493,11 +1688,12 @@ wxMBConv_iconv::wxMBConv_iconv(const wxChar *name) // first try charset with explicit bytesex info (e.g. "UCS-4LE"): wxString nameXE(nameCS); - #ifdef WORDS_BIGENDIAN + +#ifdef WORDS_BIGENDIAN nameXE += _T("BE"); - #else // little endian +#else // little endian nameXE += _T("LE"); - #endif +#endif wxLogTrace(TRACE_STRCONV, _T(" trying charset \"%s\""), nameXE.c_str()); @@ -1526,8 +1722,9 @@ wxMBConv_iconv::wxMBConv_iconv(const wxChar *name) wbufPtr = wbuf; bufPtr = buf; - res = iconv(m2w, ICONV_CHAR_CAST(&bufPtr), &insz, - (char**)&wbufPtr, &outsz); + res = iconv( + m2w, ICONV_CHAR_CAST(&bufPtr), &insz, + (char**)&wbufPtr, &outsz); if (ICONV_FAILED(res, insz)) { @@ -1593,7 +1790,7 @@ size_t wxMBConv_iconv::MB2WC(wchar_t *buf, const char *psz, size_t n) const switch ( nulLen ) { default: - return (size_t)-1; + return wxCONV_FAILED; case 1: inbuf = strlen(psz); // arguably more optimized than our version @@ -1612,8 +1809,8 @@ 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 @@ -1621,7 +1818,6 @@ size_t wxMBConv_iconv::MB2WC(wchar_t *buf, const char *psz, size_t n) const wxMutexLocker lock(wxConstCast(this, wxMBConv_iconv)->m_iconvMutex); #endif // wxUSE_THREADS - size_t outbuf = n * SIZEOF_WCHAR_T; size_t res, cres; // VS: Use these instead of psz, buf because iconv() modifies its arguments: @@ -1653,23 +1849,26 @@ size_t wxMBConv_iconv::MB2WC(wchar_t *buf, const char *psz, size_t n) const // to calculate destination buffer requirement wchar_t tbuf[8]; res = 0; - do { + + do + { bufPtr = tbuf; - outbuf = 8*SIZEOF_WCHAR_T; + outbuf = 8 * SIZEOF_WCHAR_T; cres = iconv(m2w, ICONV_CHAR_CAST(&pszPtr), &inbuf, (char**)&bufPtr, &outbuf ); - res += 8-(outbuf/SIZEOF_WCHAR_T); - } while ((cres==(size_t)-1) && (errno==E2BIG)); + res += 8 - (outbuf / SIZEOF_WCHAR_T); + } + while ((cres == (size_t)-1) && (errno == E2BIG)); } if (ICONV_FAILED(cres, inbuf)) { //VS: it is ok if iconv fails, hence trace only wxLogTrace(TRACE_STRCONV, wxT("iconv failed: %s"), wxSysErrorMsg(wxSysErrorCode())); - return (size_t)-1; + return wxCONV_FAILED; } return res; @@ -1697,6 +1896,7 @@ size_t wxMBConv_iconv::WC2MB(char *buf, const wchar_t *psz, size_t n) const tmpbuf = (wchar_t *)malloc(inbuf + SIZEOF_WCHAR_T); for ( size_t i = 0; i < inlen; i++ ) tmpbuf[n] = WC_BSWAP(psz[i]); + tmpbuf[inlen] = L'\0'; psz = tmpbuf; } @@ -1706,7 +1906,7 @@ size_t wxMBConv_iconv::WC2MB(char *buf, const wchar_t *psz, size_t n) const // have destination buffer, convert there cres = iconv( w2m, ICONV_CHAR_CAST(&psz), &inbuf, &buf, &outbuf ); - res = n-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 @@ -1716,17 +1916,20 @@ size_t wxMBConv_iconv::WC2MB(char *buf, const wchar_t *psz, size_t n) const } else { - // no destination buffer... convert using temp buffer + // no destination buffer: convert using temp buffer // to calculate destination buffer requirement char tbuf[16]; res = 0; - do { - buf = tbuf; outbuf = 16; + 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)); + } + while ((cres == (size_t)-1) && (errno == E2BIG)); } if (ms_wcNeedsSwap) @@ -1737,7 +1940,7 @@ size_t wxMBConv_iconv::WC2MB(char *buf, const wchar_t *psz, size_t n) const if (ICONV_FAILED(cres, inbuf)) { wxLogTrace(TRACE_STRCONV, wxT("iconv failed: %s"), wxSysErrorMsg(wxSysErrorCode())); - return (size_t)-1; + return wxCONV_FAILED; } return res; @@ -1758,15 +1961,15 @@ size_t wxMBConv_iconv::GetMBNulLen() const char buf[8]; // should be enough for NUL in any encoding size_t inLen = sizeof(wchar_t), outLen = WXSIZEOF(buf); - char *in = (char *)wnul; - char *out = buf; - if ( iconv(w2m, ICONV_CHAR_CAST(&in), &inLen, &out, &outLen) == (size_t)-1 ) + char *inBuff = (char *)wnul; + char *outBuff = buf; + if ( iconv(w2m, ICONV_CHAR_CAST(&inBuff), &inLen, &outBuff, &outLen) == (size_t)-1 ) { self->m_minMBCharWidth = (size_t)-1; } else // ok { - self->m_minMBCharWidth = out - buf; + self->m_minMBCharWidth = outBuff - buf; } } @@ -1797,6 +2000,13 @@ public: m_minMBCharWidth = 0; } + 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) { @@ -1811,7 +2021,7 @@ public: } #endif // wxUSE_FONTMAP - size_t MB2WC(wchar_t *buf, const char *psz, size_t n) const + virtual 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) @@ -1853,7 +2063,7 @@ public: if ( !len ) { // function totally failed - return (size_t)-1; + return wxCONV_FAILED; } // if we were really converting and didn't use MB_ERR_INVALID_CHARS, @@ -1877,7 +2087,7 @@ public: { // 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; + return wxCONV_FAILED; } } @@ -1887,7 +2097,7 @@ public: return len - 1; } - size_t WC2MB(char *buf, const wchar_t *pwz, size_t n) const + virtual size_t WC2MB(char *buf, const wchar_t *pwz, size_t n) const { /* we have a problem here: by default, WideCharToMultiByte() may @@ -1938,7 +2148,7 @@ public: if ( !len ) { // function totally failed - return (size_t)-1; + return wxCONV_FAILED; } // if we were really converting, check if we succeeded @@ -1949,17 +2159,17 @@ public: // check if the conversion failed, i.e. if any replacements // were done if ( usedDef ) - return (size_t)-1; + return wxCONV_FAILED; } else // we must resort to double tripping... { wxWCharBuffer wcBuf(n); - if ( MB2WC(wcBuf.data(), buf, n) == (size_t)-1 || + if ( MB2WC(wcBuf.data(), buf, n) == wxCONV_FAILED || 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; + return wxCONV_FAILED; } } } @@ -1989,7 +2199,8 @@ public: { default: wxLogDebug(_T("Unexpected NUL length %d"), len); - // fall through + self->m_minMBCharWidth = (size_t)-1; + break; case 0: self->m_minMBCharWidth = (size_t)-1; @@ -2006,6 +2217,8 @@ public: return m_minMBCharWidth; } + virtual wxMBConv *Clone() const { return new wxMBConv_win32(*this); } + bool IsOk() const { return m_CodePage != -1; } private: @@ -2018,17 +2231,18 @@ 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; default: - // unknown, be conseravtive by default + // unknown: be conservative by default s_isWin98Or2k = 0; + break; } wxASSERT_MSG( s_isWin98Or2k != -1, _T("should be set above") ); @@ -2081,9 +2295,9 @@ private: #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). +// 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 #include @@ -2091,12 +2305,13 @@ private: CFStringEncoding wxCFStringEncFromFontEnc(wxFontEncoding encoding) { CFStringEncoding enc = kCFStringEncodingInvalidId ; - if ( encoding == wxFONTENCODING_DEFAULT ) - { - enc = CFStringGetSystemEncoding(); - } - else switch( encoding) + + switch (encoding) { + case wxFONTENCODING_DEFAULT : + enc = CFStringGetSystemEncoding(); + break ; + case wxFONTENCODING_ISO8859_1 : enc = kCFStringEncodingISOLatin1 ; break ; @@ -2152,7 +2367,7 @@ CFStringEncoding wxCFStringEncFromFontEnc(wxFontEncoding encoding) // break ; case wxFONTENCODING_CP437 : - enc =kCFStringEncodingDOSLatinUS ; + enc = kCFStringEncodingDOSLatinUS ; break ; case wxFONTENCODING_CP850 : enc = kCFStringEncodingDOSLatin1; @@ -2164,7 +2379,7 @@ CFStringEncoding wxCFStringEncFromFontEnc(wxFontEncoding encoding) enc = kCFStringEncodingDOSCyrillic; break ; case wxFONTENCODING_CP866 : - enc =kCFStringEncodingDOSRussian ; + enc = kCFStringEncodingDOSRussian ; break ; case wxFONTENCODING_CP874 : enc = kCFStringEncodingDOSThai; @@ -2173,7 +2388,7 @@ CFStringEncoding wxCFStringEncFromFontEnc(wxFontEncoding encoding) enc = kCFStringEncodingDOSJapanese; break ; case wxFONTENCODING_CP936 : - enc =kCFStringEncodingDOSChineseSimplif ; + enc = kCFStringEncodingDOSChineseSimplif ; break ; case wxFONTENCODING_CP949 : enc = kCFStringEncodingDOSKorean; @@ -2185,10 +2400,10 @@ CFStringEncoding wxCFStringEncFromFontEnc(wxFontEncoding encoding) enc = kCFStringEncodingWindowsLatin2; break ; case wxFONTENCODING_CP1251 : - enc =kCFStringEncodingWindowsCyrillic ; + enc = kCFStringEncodingWindowsCyrillic ; break ; case wxFONTENCODING_CP1252 : - enc =kCFStringEncodingWindowsLatin1 ; + enc = kCFStringEncodingWindowsLatin1 ; break ; case wxFONTENCODING_CP1253 : enc = kCFStringEncodingWindowsGreek; @@ -2197,10 +2412,10 @@ CFStringEncoding wxCFStringEncFromFontEnc(wxFontEncoding encoding) enc = kCFStringEncodingWindowsLatin5; break ; case wxFONTENCODING_CP1255 : - enc =kCFStringEncodingWindowsHebrew ; + enc = kCFStringEncodingWindowsHebrew ; break ; case wxFONTENCODING_CP1256 : - enc =kCFStringEncodingWindowsArabic ; + enc = kCFStringEncodingWindowsArabic ; break ; case wxFONTENCODING_CP1257 : enc = kCFStringEncodingWindowsBalticRim; @@ -2338,10 +2553,12 @@ CFStringEncoding wxCFStringEncFromFontEnc(wxFontEncoding encoding) // case wxFONTENCODING_MACKEYBOARD : // enc = kCFStringEncodingMacKeyboardGlyphs ; // break ; + default : // because gcc is picky break ; - } ; + } + return enc ; } @@ -2353,6 +2570,11 @@ public: Init(CFStringGetSystemEncoding()) ; } + wxMBConv_cocoa(const wxMBConv_cocoa& conv) + { + m_encoding = conv.m_encoding; + } + #if wxUSE_FONTMAP wxMBConv_cocoa(const wxChar* name) { @@ -2365,7 +2587,7 @@ public: Init( wxCFStringEncFromFontEnc(encoding) ); } - ~wxMBConv_cocoa() + virtual ~wxMBConv_cocoa() { } @@ -2406,12 +2628,12 @@ public: CFRelease(theString); - szUniCharBuffer[nOutLength] = '\0' ; + szUniCharBuffer[nOutLength] = '\0'; #if SIZEOF_WCHAR_T == 4 - wxMBConvUTF16 converter ; - converter.MB2WC(szOut, (const char*)szUniCharBuffer , nOutSize ) ; - delete[] szUniCharBuffer; + wxMBConvUTF16 converter; + converter.MB2WC( szOut, (const char*)szUniCharBuffer, nOutSize ); + delete [] szUniCharBuffer; #endif return nOutLength; @@ -2427,9 +2649,9 @@ public: #if SIZEOF_WCHAR_T == 4 wxMBConvUTF16 converter ; - nBufSize = converter.WC2MB( NULL , szUnConv , 0 ); - szUniBuffer = new UniChar[ (nBufSize / sizeof(UniChar)) + 1] ; - converter.WC2MB( (char*) szUniBuffer , szUnConv, nBufSize + sizeof(UniChar)) ; + 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 @@ -2475,6 +2697,8 @@ public: return nRealOutSize - 1; } + virtual wxMBConv *Clone() const { return new wxMBConv_cocoa(*this); } + bool IsOk() const { return m_encoding != kCFStringEncodingInvalidId && @@ -2501,10 +2725,15 @@ public: Init(CFStringGetSystemEncoding()) ; } + wxMBConv_mac(const wxMBConv_mac& conv) + { + Init(conv.m_char_encoding); + } + #if wxUSE_FONTMAP wxMBConv_mac(const wxChar* name) { - Init( wxMacGetSystemEncFromFontEnc(wxFontMapperBase::Get()->CharsetToEncoding(name, false) ) ) ; + Init( wxMacGetSystemEncFromFontEnc( wxFontMapperBase::Get()->CharsetToEncoding(name, false) ) ); } #endif @@ -2513,61 +2742,80 @@ 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_unicode_encoding = CreateTextEncoding(kTextEncodingUnicodeDefault,0,kUnicode16BitFormat) ; + 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) ; + 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) ; + // 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 - status = TECConvertText(m_MB2WC_converter, (ConstTextPtr) psz , byteInLen, &byteInLen, - (TextPtr) ubuf , byteBufferLen, &byteOutLen); + + 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 ; wxMBConvUTF16 converter ; - res = converter.MB2WC( (buf ? buf : tbuf) , (const char*)ubuf , n ) ; + res = converter.MB2WC( (buf ? buf : tbuf), (const char*)ubuf, n ) ; free( ubuf ) ; #else res = byteOutLen / sizeof( UniChar ) ; #endif + if ( buf == NULL ) free(tbuf) ; @@ -2579,6 +2827,7 @@ 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 ; @@ -2587,27 +2836,32 @@ public: if (buf == NULL) { - //apple specs say at least 32 - n = wxMax( 32 , ((byteInLen / SIZEOF_WCHAR_T) * 8) + SIZEOF_WCHAR_T ); + // 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 ) ; + size_t unicharlen = converter.WC2MB( NULL, psz, 0 ) ; byteInLen = unicharlen ; ubuf = (UniChar*) malloc( byteInLen + 2 ) ; - converter.WC2MB( (char*) ubuf , psz, unicharlen + 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); + + 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) ; @@ -2620,30 +2874,208 @@ public: //of bogus characters wxWCharBuffer wcBuf(n); size_t pszlen = wxWcslen(psz); - if ( MB2WC(wcBuf.data(), buf, n) == (size_t)-1 || + if ( MB2WC(wcBuf.data(), buf, n) == wxCONV_FAILED || wxWcslen(wcBuf) != pszlen || memcmp(wcBuf, psz, pszlen * sizeof(wchar_t)) != 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; + return wxCONV_FAILED; } } return res ; } + 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 ; + 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) // ============================================================================ @@ -2686,8 +3118,8 @@ public: size_t inbuf = strlen(psz); if (buf) { - if (!m2w.Convert(psz,buf)) - return (size_t)-1; + if (!m2w.Convert(psz, buf)) + return wxCONV_FAILED; } return inbuf; } @@ -2697,8 +3129,8 @@ public: const size_t inbuf = wxWcslen(psz); if (buf) { - if (!w2m.Convert(psz,buf)) - return (size_t)-1; + if (!w2m.Convert(psz, buf)) + return wxCONV_FAILED; } return inbuf; @@ -2721,6 +3153,8 @@ public: } } + virtual wxMBConv *Clone() const { return new wxMBConv_wxwin(m_enc); } + bool IsOk() const { return m_ok; } public: @@ -2743,6 +3177,7 @@ WXDLLIMPEXP_BASE wxMBConv* new_wxMBConv_wxwin( const wxChar* name ) delete result; return 0; } + return result; } @@ -2832,7 +3267,6 @@ void wxCSConv::SetName(const wxChar *charset) } #if wxUSE_FONTMAP -#include "wx/hashmap.h" WX_DECLARE_HASH_MAP( wxFontEncoding, wxString, wxIntegerHash, wxIntegerEqual, wxEncodingNameCache ); @@ -2874,7 +3308,9 @@ wxMBConv *wxCSConv::DoCreate() const #endif // !wxUSE_FONTMAP { wxString name(m_name); +#if wxUSE_FONTMAP wxFontEncoding encoding(m_encoding); +#endif if ( !name.empty() ) { @@ -2905,20 +3341,26 @@ 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; + wxMBConv_iconv *conv = new wxMBConv_iconv(*names); + 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 } @@ -2938,13 +3380,13 @@ wxMBConv *wxCSConv::DoCreate() const #endif } #endif // wxHAVE_WIN32_MB2WC + #if defined(__WXMAC__) { // leave UTF16 and UTF32 to the built-ins of wx if ( m_name || ( m_encoding < wxFONTENCODING_UTF16BE || ( m_encoding >= wxFONTENCODING_MACMIN && m_encoding <= wxFONTENCODING_MACMAX ) ) ) { - #if wxUSE_FONTMAP wxMBConv_mac *conv = m_name ? new wxMBConv_mac(m_name) : new wxMBConv_mac(m_encoding); @@ -2958,17 +3400,18 @@ wxMBConv *wxCSConv::DoCreate() const } } #endif + #if defined(__WXCOCOA__) { if ( m_name || ( m_encoding <= wxFONTENCODING_UTF16 ) ) { - #if wxUSE_FONTMAP wxMBConv_cocoa *conv = m_name ? new wxMBConv_cocoa(m_name) : new wxMBConv_cocoa(m_encoding); #else wxMBConv_cocoa *conv = new wxMBConv_cocoa(m_encoding); #endif + if ( conv->IsOk() ) return conv; @@ -3010,7 +3453,7 @@ wxMBConv *wxCSConv::DoCreate() const default: // nothing to do but put here to suppress gcc warnings - ; + break; } // step (3) @@ -3028,10 +3471,10 @@ 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) { @@ -3042,9 +3485,10 @@ wxMBConv *wxCSConv::DoCreate() const #if wxUSE_FONTMAP wxFontMapperBase::GetEncodingDescription(m_encoding).c_str() #else // !wxUSE_FONTMAP - wxString::Format(_("encoding %s"), m_encoding).c_str() + wxString::Format(_("encoding %i"), m_encoding).c_str() #endif // wxUSE_FONTMAP/!wxUSE_FONTMAP ); + alreadyLoggingError = false; } @@ -3057,14 +3501,17 @@ 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 ) { +#if wxUSE_INTL self->m_name = wxStrdup(wxLocale::GetSystemEncodingName()); - } +#else + // fallback to some reasonable default: + self->m_encoding = wxFONTENCODING_ISO8859_1; #endif // wxUSE_INTL + } self->m_convReal = DoCreate(); self->m_deferred = false; @@ -3104,7 +3551,8 @@ size_t wxCSConv::WC2MB(char *buf, const wchar_t *psz, size_t n) const for (size_t c = 0; c <= len; c++) { if (psz[c] > 0xFF) - return (size_t)-1; + return wxCONV_FAILED; + buf[c] = (char)psz[c]; } } @@ -3113,7 +3561,7 @@ size_t wxCSConv::WC2MB(char *buf, const wchar_t *psz, size_t n) const for (size_t c = 0; c <= len; c++) { if (psz[c] > 0xFF) - return (size_t)-1; + return wxCONV_FAILED; } } @@ -3148,21 +3596,27 @@ static wxCSConv wxConvLocalObj(wxFONTENCODING_SYSTEM); static wxCSConv wxConvISO8859_1Obj(wxFONTENCODING_ISO8859_1); static wxMBConvUTF7 wxConvUTF7Obj; static wxMBConvUTF8 wxConvUTF8Obj; - +#if defined(__WXMAC__) && defined(TARGET_CARBON) +static wxMBConv_macUTF8D wxConvMacUTF8DObj; +#endif 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 = & #ifdef __WXOSX__ +#if defined(__WXMAC__) && defined(TARGET_CARBON) + wxConvMacUTF8DObj; +#else wxConvUTF8Obj; +#endif #else wxConvLibcObj; #endif - #else // !wxUSE_WCHAR_T // stand-ins in absence of wchar_t