X-Git-Url: https://git.saurik.com/wxWidgets.git/blobdiff_plain/5ce0e4ac0617023b0eeabd836df338c77e5f478d..73fe67bd60b57f95dc63809f7843ed2a15928436:/src/common/strconv.cpp diff --git a/src/common/strconv.cpp b/src/common/strconv.cpp index fdb714185f..4b2a491e13 100644 --- a/src/common/strconv.cpp +++ b/src/common/strconv.cpp @@ -1,12 +1,13 @@ ///////////////////////////////////////////////////////////////////////////// // Name: strconv.cpp // Purpose: Unicode conversion classes -// Author: Ove Kaaven, Robert Roebling, Vadim Zeitlin +// Author: Ove Kaaven, Robert Roebling, Vadim Zeitlin, Vaclav Slavik // Modified by: // Created: 29/01/98 // RCS-ID: $Id$ -// Copyright: (c) 1999 Ove Kaaven, Robert Roebling, Vadim Zeitlin -// Licence: wxWindows license +// Copyright: (c) 1999 Ove Kaaven, Robert Roebling, Vaclav Slavik +// (c) 2000-2003 Vadim Zeitlin +// Licence: wxWindows licence ///////////////////////////////////////////////////////////////////////////// // ============================================================================ @@ -17,7 +18,7 @@ // headers // ---------------------------------------------------------------------------- -#ifdef __GNUG__ +#if defined(__GNUG__) && !defined(NO_GCC_PRAGMA) #pragma implementation "strconv.h" #endif @@ -28,208 +29,202 @@ #pragma hdrstop #endif +#ifndef WX_PRECOMP + #include "wx/intl.h" + #include "wx/log.h" +#endif // WX_PRECOMP + +#include "wx/strconv.h" + +#if wxUSE_WCHAR_T + #ifdef __WXMSW__ - #include "wx/msw/private.h" + #include "wx/msw/private.h" + #include "wx/msw/missing.h" #endif +#ifndef __WXWINCE__ #include +#endif + #include #include #include +#if defined(__WIN32__) && !defined(__WXMICROWIN__) + #define wxHAVE_WIN32_MB2WC +#endif // __WIN32__ but !__WXMICROWIN__ + +// ---------------------------------------------------------------------------- +// headers +// ---------------------------------------------------------------------------- + #ifdef __SALFORDC__ - #include + #include #endif -#ifdef HAVE_ICONV_H - #include -#endif -#ifdef HAVE_LANGINFO_H - #include +#ifdef HAVE_ICONV + #include #endif -#ifdef __WXMSW__ - #include -#endif +#include "wx/encconv.h" +#include "wx/fontmap.h" -#include "wx/debug.h" -#include "wx/strconv.h" -#include "wx/intl.h" -#include "wx/log.h" +#ifdef __WXMAC__ +#include +#include +#include -#if defined(WORDS_BIGENDIAN) || defined(__STDC_ISO_10646__) -#define BSWAP_UCS4(str, len) -#define BSWAP_UCS2(str, len) -#else -#define BSWAP_UCS4(str, len) { unsigned _c; for (_c=0; _c=0x110000) { - return (size_t)-1; - } else { - if (output) { - *output++ = (input >> 10)+0xd7c0; - *output++ = (input&0x3ff)+0xdc00; + if (input<=0xffff) + { + if (output) + *output = (wxUint16) input; + return 1; + } + else if (input>=0x110000) + { + return (size_t)-1; + } + else + { + if (output) + { + *output++ = (wxUint16) ((input >> 10)+0xd7c0); + *output = (wxUint16) ((input&0x3ff)+0xdc00); + } + return 2; } - return 2; - } } -static size_t decode_utf16(wxUint16*input,wxUint32&output) +static size_t decode_utf16(const wxUint16* input, wxUint32& output) { - if ((*input<0xd800) || (*input>0xdfff)) { - output = *input; - return 1; - } else - if ((input[1]<0xdc00) || (input[1]>=0xdfff)) { - output = *input; - return (size_t)-1; - } else { - output = ((input[0] - 0xd7c0) << 10) + (input[1] - 0xdc00); - return 2; - } + if ((*input<0xd800) || (*input>0xdfff)) + { + output = *input; + return 1; + } + else if ((input[1]<0xdc00) || (input[1]>=0xdfff)) + { + output = *input; + return (size_t)-1; + } + else + { + output = ((input[0] - 0xd7c0) << 10) + (input[1] - 0xdc00); + return 2; + } } -#endif // WC_UTF16 // ---------------------------------------------------------------------------- // wxMBConv // ---------------------------------------------------------------------------- -WXDLLEXPORT_DATA(wxMBConv) wxConvLibc; - -size_t wxMBConv::MB2WC(wchar_t *buf, const char *psz, size_t n) const -{ - return wxMB2WC(buf, psz, n); -} - -size_t wxMBConv::WC2MB(char *buf, const wchar_t *psz, size_t n) const +wxMBConv::~wxMBConv() { - return wxWC2MB(buf, psz, n); + // nothing to do here } const wxWCharBuffer wxMBConv::cMB2WC(const char *psz) const { - if (psz) + if ( psz ) { - size_t nLen = MB2WC((wchar_t *) NULL, psz, 0); - if (nLen == (size_t)-1) - return wxWCharBuffer((wchar_t *) NULL); - wxWCharBuffer buf(nLen); - MB2WC((wchar_t *)(const wchar_t *) buf, psz, nLen); - return buf; + // calculate the length of the buffer needed first + size_t nLen = MB2WC(NULL, psz, 0); + if ( nLen != (size_t)-1 ) + { + // now do the actual conversion + wxWCharBuffer buf(nLen); + MB2WC(buf.data(), psz, nLen + 1); // with the trailing NUL + + return buf; + } } - else - return wxWCharBuffer((wchar_t *) NULL); + + wxWCharBuffer buf((wchar_t *)NULL); + + return buf; } -const wxCharBuffer wxMBConv::cWC2MB(const wchar_t *psz) const +const wxCharBuffer wxMBConv::cWC2MB(const wchar_t *pwz) const { - if (psz) + if ( pwz ) { - size_t nLen = WC2MB((char *) NULL, psz, 0); - if (nLen == (size_t)-1) - return wxCharBuffer((char *) NULL); - wxCharBuffer buf(nLen); - WC2MB((char *)(const char *) buf, psz, nLen); - return buf; - } - else - return wxCharBuffer((char *) NULL); -} - -// ---------------------------------------------------------------------------- -// standard file conversion -// ---------------------------------------------------------------------------- + size_t nLen = WC2MB(NULL, pwz, 0); + if ( nLen != (size_t)-1 ) + { + wxCharBuffer buf(nLen+3); // space for a wxUint32 trailing zero + WC2MB(buf.data(), pwz, nLen + 4); -WXDLLEXPORT_DATA(wxMBConvFile) wxConvFile; + return buf; + } + } -// just use the libc conversion for now -size_t wxMBConvFile::MB2WC(wchar_t *buf, const char *psz, size_t n) const -{ - return wxMB2WC(buf, psz, n); -} + wxCharBuffer buf((char *)NULL); -size_t wxMBConvFile::WC2MB(char *buf, const wchar_t *psz, size_t n) const -{ - return wxWC2MB(buf, psz, n); + return buf; } // ---------------------------------------------------------------------------- -// standard gdk conversion +// wxMBConvLibc // ---------------------------------------------------------------------------- -#ifdef __WXGTK12__ - -WXDLLEXPORT_DATA(wxMBConvGdk) wxConvGdk; - -#include - -size_t wxMBConvGdk::MB2WC(wchar_t *buf, const char *psz, size_t n) const +size_t wxMBConvLibc::MB2WC(wchar_t *buf, const char *psz, size_t n) const { - if (buf) { - return gdk_mbstowcs((GdkWChar *)buf, psz, n); - } else { - GdkWChar *nbuf = new GdkWChar[n=strlen(psz)]; - size_t len = gdk_mbstowcs(nbuf, psz, n); - delete [] nbuf; - return len; - } + return wxMB2WC(buf, psz, n); } -size_t wxMBConvGdk::WC2MB(char *buf, const wchar_t *psz, size_t n) const +size_t wxMBConvLibc::WC2MB(char *buf, const wchar_t *psz, size_t n) const { - char *mbstr = gdk_wcstombs((GdkWChar *)psz); - size_t len = mbstr ? strlen(mbstr) : 0; - if (buf) { - if (len > n) len = n; - memcpy(buf, psz, len); - if (len < n) buf[len] = 0; - } - return len; + return wxWC2MB(buf, psz, n); } -#endif // GTK > 1.0 - // ---------------------------------------------------------------------------- // UTF-7 // ---------------------------------------------------------------------------- -WXDLLEXPORT_DATA(wxMBConvUTF7) wxConvUTF7; - #if 0 static char utf7_setD[]="ABCDEFGHIJKLMNOPQRSTUVWXYZ" "abcdefghijklmnopqrstuvwxyz" @@ -259,430 +254,1296 @@ size_t wxMBConvUTF7::WC2MB(char * WXUNUSED(buf), // UTF-8 // ---------------------------------------------------------------------------- -WXDLLEXPORT_DATA(wxMBConvUTF8) wxConvUTF8; - -static wxUint32 utf8_max[]={0x7f,0x7ff,0xffff,0x1fffff,0x3ffffff,0x7fffffff,0xffffffff}; +static wxUint32 utf8_max[]= + { 0x7f, 0x7ff, 0xffff, 0x1fffff, 0x3ffffff, 0x7fffffff, 0xffffffff }; size_t wxMBConvUTF8::MB2WC(wchar_t *buf, const char *psz, size_t n) const { - size_t len = 0; - - while (*psz && ((!buf) || (len>cnt); - while (cnt--) { - cc = *psz++; - if ((cc&0xC0)!=0x80) { - // invalid UTF-8 sequence - return (size_t)-1; - } - res=(res<<6)|(cc&0x3f); - } - if (res<=utf8_max[ocnt]) { - // illegal UTF-8 encoding - return (size_t)-1; + size_t len = 0; + + while (*psz && ((!buf) || (len < n))) + { + unsigned char cc = *psz++, fc = cc; + unsigned cnt; + for (cnt = 0; fc & 0x80; cnt++) + fc <<= 1; + if (!cnt) + { + // plain ASCII char + if (buf) + *buf++ = cc; + len++; } + else + { + cnt--; + if (!cnt) + { + // invalid UTF-8 sequence + return (size_t)-1; + } + else + { + unsigned ocnt = cnt - 1; + wxUint32 res = cc & (0x3f >> cnt); + while (cnt--) + { + cc = *psz++; + if ((cc & 0xC0) != 0x80) + { + // invalid UTF-8 sequence + return (size_t)-1; + } + res = (res << 6) | (cc & 0x3f); + } + if (res <= utf8_max[ocnt]) + { + // illegal UTF-8 encoding + return (size_t)-1; + } #ifdef WC_UTF16 - size_t pa = encode_utf16(res, buf); - if (pa == (size_t)-1) - return (size_t)-1; - if (buf) buf+=pa; - len+=pa; -#else - if (buf) *buf++=res; - len++; -#endif - } + // cast is ok because wchar_t == wxUuint16 if WC_UTF16 + size_t pa = encode_utf16(res, (wxUint16 *)buf); + if (pa == (size_t)-1) + return (size_t)-1; + if (buf) + buf += pa; + len += pa; +#else // !WC_UTF16 + if (buf) + *buf++ = res; + len++; +#endif // WC_UTF16/!WC_UTF16 + } + } } - } - if (buf && (lenutf8_max[cnt]; cnt++); - if (!cnt) { - // plain ASCII char - if (buf) *buf++=cc; - len++; - } else { - len+=cnt+1; - if (buf) { - *buf++=(-128>>cnt)|((cc>>(cnt*6))&(0x3f>>cnt)); - while (cnt--) - *buf++=0x80|((cc>>(cnt*6))&0x3f); - } - } - } - if (buf && (len utf8_max[cnt]; cnt++) {} + if (!cnt) + { + // plain ASCII char + if (buf) + *buf++ = (char) cc; + len++; + } + + else + { + len += cnt + 1; + if (buf) + { + *buf++ = (char) ((-128 >> cnt) | ((cc >> (cnt * 6)) & (0x3f >> cnt))); + while (cnt--) + *buf++ = (char) (0x80 | ((cc >> (cnt * 6)) & 0x3f)); + } + } + } + + if (buf && (len wxConvUTF8) -// - move wxEncodingConverter meat in here +#ifdef WC_UTF16 -#ifdef __WIN32__ -#include "wx/msw/registry.h" -// this should work if M$ Internet Exploiter is installed -static long CharsetToCodepage(const wxChar *name) +// copy 16bit MB to 16bit String +size_t wxMBConvUTF16straight::MB2WC(wchar_t *buf, const char *psz, size_t n) const { - if (!name) - return GetACP(); - - long CP=-1; - - wxString cn(name); - do { - wxString path(wxT("MIME\\Database\\Charset\\")); - path += cn; - wxRegKey key(wxRegKey::HKCR, path); + size_t len=0; - if (!key.Exists()) continue; - - // two cases: either there's an AliasForCharset string, - // or there are Codepage and InternetEncoding dwords. - // The InternetEncoding gives us the actual encoding, - // the Codepage just says which Windows character set to - // use when displaying the data. - if (key.HasValue(wxT("InternetEncoding")) && - key.QueryValue(wxT("InternetEncoding"), &CP)) break; - - // no encoding, see if it's an alias - if (!key.HasValue(wxT("AliasForCharset")) || - !key.QueryValue(wxT("AliasForCharset"), cn)) break; - } while (1); - - return CP; + while (*(wxUint16*)psz && (!buf || len < n)) + { + if (buf) + *buf++ = *(wxUint16*)psz; + len++; + + psz += sizeof(wxUint16); + } + if (buf && lenMB2WC(buf,psz,n) : (size_t)-1; } - - size_t WC2MB(char*buf, const wchar_t*psz, size_t n) - { return work ? work->WC2MB(buf,psz,n) : (size_t)-1; } - - bool usable() - { return work!=NULL; } -public: - wxMBConv*work; -}; + size_t len=0; + while (*psz && (!buf || len < n)) + { + if (buf) + { + *(wxUint16*)buf = *psz; + buf += sizeof(wxUint16); + } + len += sizeof(wxUint16); + psz++; + } + if (buf && len<=n-sizeof(wxUint16)) *(wxUint16*)buf=0; -#ifdef HAVE_ICONV_H + return len; +} -// VS: glibc 2.1.3 is broken in that iconv() conversion to/from UCS4 fails with E2BIG -// if output buffer is _exactly_ as big as needed. Such case is (unless there's -// yet another bug in glibc) the only case when iconv() returns with (size_t)-1 -// (which means error) and says there are 0 bytes left in the input buffer -- -// when _real_ error occurs, bytes-left-in-input buffer is non-zero. Hence, -// this alternative test for iconv() failure. -// [This bug does not appear in glibc 2.2.] -#if defined(__GLIBC__) && __GLIBC__ == 2 && __GLIBC_MINOR__ <= 1 -#define ICONV_FAILED(cres, bufLeft) ((cres == (size_t)-1) && \ - (errno != E2BIG || bufLeft != 0)) -#else -#define ICONV_FAILED(cres, bufLeft) (cres == (size_t)-1) -#endif -class IC_CharSet : public wxCharacterSet +// swap 16bit MB to 16bit String +size_t wxMBConvUTF16swap::MB2WC(wchar_t *buf, const char *psz, size_t n) const { -public: - IC_CharSet(const wxChar*name) - : wxCharacterSet(name) - { - m2w = iconv_open(WC_NAME, wxConvLibc.cWX2MB(cname)); - w2m = iconv_open(wxConvLibc.cWX2MB(cname), WC_NAME); - } - - ~IC_CharSet() - { - if ( m2w != (iconv_t)-1 ) - iconv_close(m2w); - if ( w2m != (iconv_t)-1 ) - iconv_close(w2m); - } - - size_t MB2WC(wchar_t *buf, const char *psz, size_t n) - { - size_t inbuf = strlen(psz); - size_t outbuf = n * SIZEOF_WCHAR_T; - size_t res, cres; - // VS: Use these instead of psz, buf because iconv() modifies its arguments: - wchar_t *bufPtr = buf; - const char *pszPtr = psz; + size_t len=0; + while (*(wxUint16*)psz && (!buf || len < n)) + { if (buf) { - // have destination buffer, convert there -#ifdef WX_ICONV_TAKES_CHAR - cres = iconv(m2w, (char**)&pszPtr, &inbuf, (char**)&bufPtr, &outbuf); -#else - cres = iconv(m2w, &pszPtr, &inbuf, (char**)&bufPtr, &outbuf); -#endif - res = n - (outbuf / SIZEOF_WCHAR_T); - // convert to native endianness -#ifdef WC_NEED_BSWAP - WC_BSWAP(buf /* _not_ bufPtr */, res) -#endif - } - else - { - // no destination buffer... convert using temp buffer - // to calculate destination buffer requirement - wchar_t tbuf[8]; - res = 0; - do { - bufPtr = tbuf; outbuf = 8*SIZEOF_WCHAR_T; -#ifdef WX_ICONV_TAKES_CHAR - cres = iconv( m2w, (char**)&pszPtr, &inbuf, (char**)&bufPtr, &outbuf ); -#else - cres = iconv( m2w, &pszPtr, &inbuf, (char**)&bufPtr, &outbuf ); -#endif - res += 8-(outbuf/SIZEOF_WCHAR_T); - } while ((cres==(size_t)-1) && (errno==E2BIG)); + ((char *)buf)[0] = psz[1]; + ((char *)buf)[1] = psz[0]; + buf++; } - - if (ICONV_FAILED(cres, inbuf)) - return (size_t)-1; - - return res; + len++; + psz += sizeof(wxUint16); } - - size_t WC2MB(char*buf, const wchar_t*psz, size_t n) - { -#if defined(__BORLANDC__) && (__BORLANDC__ > 0x530) - size_t inbuf = std::wcslen(psz) * SIZEOF_WCHAR_T; -#else - size_t inbuf = ::wcslen(psz) * SIZEOF_WCHAR_T; -#endif - size_t outbuf = n; - size_t res, cres; + if (buf && lenCharsetToEncoding(name, FALSE); - m2w.Init(enc, wxFONTENCODING_UNICODE); - w2m.Init(wxFONTENCODING_UNICODE, enc); - } - - size_t MB2WC(wchar_t*buf, const char*psz, size_t n) + wxUint16 cc[2]; + size_t pa=encode_utf16(*psz, cc); + + if (pa == (size_t)-1) + return pa; + + if (buf) + { + *(wxUint16*)buf = cc[0]; + buf += sizeof(wxUint16); + if (pa > 1) + { + *(wxUint16*)buf = cc[1]; + buf += sizeof(wxUint16); + } + } + + len += pa*sizeof(wxUint16); + psz++; + } + if (buf && len<=n-sizeof(wxUint16)) *(wxUint16*)buf=0; + + return len; +} + + +// swap 16bit MB to 32bit String +size_t wxMBConvUTF16swap::MB2WC(wchar_t *buf, const char *psz, size_t n) const +{ + size_t len=0; + + while (*(wxUint16*)psz && (!buf || len < n)) + { + wxUint32 cc; + char tmp[4]; + tmp[0]=psz[1]; tmp[1]=psz[0]; + tmp[2]=psz[3]; tmp[3]=psz[2]; + + size_t pa=decode_utf16((wxUint16*)tmp, cc); + if (pa == (size_t)-1) + return pa; + + if (buf) + *buf++ = cc; + + len++; + psz += pa * sizeof(wxUint16); + } + if (buf && len 1) + { + *buf++ = ((char*)cc)[3]; + *buf++ = ((char*)cc)[2]; + } + } + + len += pa*sizeof(wxUint16); + psz++; + } + if (buf && len<=n-sizeof(wxUint16)) *(wxUint16*)buf=0; + + return len; +} + +#endif // WC_UTF16 + + +// ---------------------------------------------------------------------------- +// UTF-32 +// ---------------------------------------------------------------------------- + +#ifdef WORDS_BIGENDIAN +#define wxMBConvUTF32straight wxMBConvUTF32BE +#define wxMBConvUTF32swap wxMBConvUTF32LE +#else +#define wxMBConvUTF32swap wxMBConvUTF32BE +#define wxMBConvUTF32straight wxMBConvUTF32LE +#endif + + +WXDLLIMPEXP_DATA_BASE(wxMBConvUTF32LE) wxConvUTF32LE; +WXDLLIMPEXP_DATA_BASE(wxMBConvUTF32BE) wxConvUTF32BE; + + +#ifdef WC_UTF16 + +// copy 32bit MB to 16bit String +size_t wxMBConvUTF32straight::MB2WC(wchar_t *buf, const char *psz, size_t n) const +{ + size_t len=0; + + while (*(wxUint32*)psz && (!buf || len < n)) + { + wxUint16 cc[2]; + + size_t pa=encode_utf16(*(wxUint32*)psz, cc); + if (pa == (size_t)-1) + return pa; + + if (buf) + { + *buf++ = cc[0]; + if (pa > 1) + *buf++ = cc[1]; + } + len += pa; + psz += sizeof(wxUint32); + } + if (buf && len 1) + *buf++ = cc[1]; + } + len += pa; + psz += sizeof(wxUint32); + } + + if (buf && len= 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 + +// ============================================================================ +// Mac conversion classes +// ============================================================================ + +#if defined(__WXMAC__) && defined(TARGET_CARBON) + +class wxMBConv_mac : public wxMBConv +{ +public: + wxMBConv_mac() + { + Init(CFStringGetSystemEncoding()) ; + } + + wxMBConv_mac(const wxChar* name) + { + Init( wxMacGetSystemEncFromFontEnc(wxFontMapper::Get()->CharsetToEncoding(name, FALSE) ) ) ; } + + wxMBConv_mac(wxFontEncoding encoding) + { + Init( wxMacGetSystemEncFromFontEnc(encoding) ); + } + + ~wxMBConv_mac() + { + OSStatus status = noErr ; + status = TECDisposeConverter(m_MB2WC_converter); + status = TECDisposeConverter(m_WC2MB_converter); + } + - size_t WC2MB(char*buf, const wchar_t*psz, size_t n) + void Init( TextEncodingBase encoding) + { + OSStatus status = noErr ; + m_char_encoding = encoding ; + m_unicode_encoding = CreateTextEncoding(kTextEncodingUnicodeDefault,0,kUnicode16BitFormat) ; + + status = TECCreateConverter(&m_MB2WC_converter, + m_char_encoding, + m_unicode_encoding); + status = TECCreateConverter(&m_WC2MB_converter, + m_unicode_encoding, + m_char_encoding); + } + + size_t MB2WC(wchar_t *buf, const char *psz, size_t n) const { -#if defined(__BORLANDC__) && (__BORLANDC__ > 0x530) - size_t inbuf = std::wcslen(psz); + OSStatus status = noErr ; + ByteCount byteOutLen ; + ByteCount byteInLen = strlen(psz) ; + wchar_t *tbuf = NULL ; + UniChar* ubuf = NULL ; + size_t res = 0 ; + + if (buf == NULL) + { + n = byteInLen ; + tbuf = (wchar_t*) malloc( n * SIZEOF_WCHAR_T) ; + } + ByteCount byteBufferLen = n * sizeof( UniChar ) ; +#if SIZEOF_WCHAR_T == 4 + ubuf = (UniChar*) malloc( byteBufferLen + 2 ) ; +#else + ubuf = (UniChar*) (buf ? buf : tbuf) ; +#endif + status = TECConvertText(m_MB2WC_converter, (ConstTextPtr) psz , byteInLen, &byteInLen, + (TextPtr) ubuf , byteBufferLen, &byteOutLen); +#if SIZEOF_WCHAR_T == 4 + // we have to terminate here, because n might be larger for the trailing zero, and if UniChar + // is not properly terminated we get random characters at the end + ubuf[byteOutLen / sizeof( UniChar ) ] = 0 ; + wxMBConvUTF16BE converter ; + res = converter.MB2WC( (buf ? buf : tbuf) , (const char*)ubuf , n ) ; + free( ubuf ) ; +#else + res = byteOutLen / sizeof( UniChar ) ; +#endif + if ( buf == NULL ) + free(tbuf) ; + + if ( buf && res < n) + buf[res] = 0; + + return res ; + } + + size_t WC2MB(char *buf, const wchar_t *psz, size_t n) const + { + OSStatus status = noErr ; + ByteCount byteOutLen ; + ByteCount byteInLen = wxWcslen(psz) * SIZEOF_WCHAR_T ; + + char *tbuf = NULL ; + + if (buf == NULL) + { + // worst case + n = byteInLen * 2 ; + tbuf = (char*) malloc( n ) ; + } + + ByteCount byteBufferLen = n ; + UniChar* ubuf = NULL ; +#if SIZEOF_WCHAR_T == 4 + wxMBConvUTF16BE converter ; + size_t unicharlen = converter.WC2MB( NULL , psz , 0 ) ; + byteInLen = unicharlen ; + ubuf = (UniChar*) malloc( byteInLen + 2 ) ; + converter.WC2MB( (char*) ubuf , psz, unicharlen + 2 ) ; #else - size_t inbuf = ::wcslen(psz); + ubuf = (UniChar*) psz ; +#endif + status = TECConvertText(m_WC2MB_converter, (ConstTextPtr) ubuf , byteInLen, &byteInLen, + (TextPtr) (buf ? buf : tbuf) , byteBufferLen, &byteOutLen); +#if SIZEOF_WCHAR_T == 4 + free( ubuf ) ; #endif + if ( buf == NULL ) + free(tbuf) ; + + size_t res = byteOutLen ; + if ( buf && res < n) + buf[res] = 0; + + return res ; + } + + bool IsOk() const + { return m_MB2WC_converter != NULL && m_WC2MB_converter != NULL ; } + +private: + TECObjectRef m_MB2WC_converter ; + TECObjectRef m_WC2MB_converter ; + + TextEncodingBase m_char_encoding ; + TextEncodingBase m_unicode_encoding ; +}; + +#endif // defined(__WXMAC__) && defined(TARGET_CARBON) + +// ============================================================================ +// wxEncodingConverter based conversion classes +// ============================================================================ + +#if wxUSE_FONTMAP + +class wxMBConv_wxwin : public wxMBConv +{ +private: + void Init() + { + m_ok = m2w.Init(m_enc, wxFONTENCODING_UNICODE) && + w2m.Init(wxFONTENCODING_UNICODE, m_enc); + } + +public: + // temporarily just use wxEncodingConverter stuff, + // so that it works while a better implementation is built + wxMBConv_wxwin(const wxChar* name) + { + if (name) + m_enc = wxFontMapper::Get()->CharsetToEncoding(name, false); + else + m_enc = wxFONTENCODING_SYSTEM; + + Init(); + } + + wxMBConv_wxwin(wxFontEncoding enc) + { + m_enc = enc; + + Init(); + } + + size_t MB2WC(wchar_t *buf, const char *psz, size_t WXUNUSED(n)) const + { + size_t inbuf = strlen(psz); + if (buf) + m2w.Convert(psz,buf); + return inbuf; + } + + size_t WC2MB(char *buf, const wchar_t *psz, size_t WXUNUSED(n)) const + { + const size_t inbuf = wxWcslen(psz); if (buf) w2m.Convert(psz,buf); - + return inbuf; } - - bool usable() - { return (enc!=wxFONTENCODING_SYSTEM) && (enc!=wxFONTENCODING_DEFAULT); } + + bool IsOk() const { return m_ok; } public: - wxFontEncoding enc; + wxFontEncoding m_enc; wxEncodingConverter m2w, w2m; + + // were we initialized successfully? + bool m_ok; + + DECLARE_NO_COPY_CLASS(wxMBConv_wxwin) }; -static wxCharacterSet *wxGetCharacterSet(const wxChar *name) +#endif // wxUSE_FONTMAP + +// ============================================================================ +// wxCSConv implementation +// ============================================================================ + +void wxCSConv::Init() +{ + m_name = NULL; + m_convReal = NULL; + m_deferred = true; +} + +wxCSConv::wxCSConv(const wxChar *charset) { - wxCharacterSet *cset = NULL; - if (name) + Init(); + + if ( charset ) { - if (!wxStricmp(name, wxT("UTF8")) || !wxStricmp(name, wxT("UTF-8"))) - { - cset = new ID_CharSet(name, &wxConvUTF8); - } - else - { -#ifdef HAVE_ICONV_H - cset = new IC_CharSet(name); // may not take NULL -#endif - } + SetName(charset); } - - if (cset && cset->usable()) return cset; - if (cset) delete cset; - cset = NULL; -#ifdef __WIN32__ - cset = new CP_CharSet(name); // may take NULL - if (cset->usable()) return cset; -#endif - if (cset) delete cset; - cset = new EC_CharSet(name); - if (cset->usable()) return cset; - delete cset; - wxLogError(_("Unknown encoding '%s'!"), name); - return NULL; + + m_encoding = wxFONTENCODING_SYSTEM; } -wxCSConv::wxCSConv(const wxChar *charset) +wxCSConv::wxCSConv(wxFontEncoding encoding) { - m_name = (wxChar *) NULL; - m_cset = (wxCharacterSet *) NULL; - m_deferred = TRUE; - SetName(charset); + if ( encoding == wxFONTENCODING_MAX || encoding == wxFONTENCODING_DEFAULT ) + { + wxFAIL_MSG( _T("invalid encoding value in wxCSConv ctor") ); + + encoding = wxFONTENCODING_SYSTEM; + } + + Init(); + + m_encoding = encoding; } wxCSConv::~wxCSConv() { - if (m_name) free(m_name); - if (m_cset) delete m_cset; + Clear(); +} + +wxCSConv::wxCSConv(const wxCSConv& conv) + : wxMBConv() +{ + Init(); + + SetName(conv.m_name); + m_encoding = conv.m_encoding; +} + +wxCSConv& wxCSConv::operator=(const wxCSConv& conv) +{ + Clear(); + + SetName(conv.m_name); + m_encoding = conv.m_encoding; + + return *this; +} + +void wxCSConv::Clear() +{ + free(m_name); + delete m_convReal; + + m_name = NULL; + m_convReal = NULL; } void wxCSConv::SetName(const wxChar *charset) @@ -690,137 +1551,249 @@ void wxCSConv::SetName(const wxChar *charset) if (charset) { m_name = wxStrdup(charset); - m_deferred = TRUE; + m_deferred = true; } } -void wxCSConv::LoadNow() +wxMBConv *wxCSConv::DoCreate() const { - if (m_deferred) + // check for the special case of ASCII or ISO8859-1 charset: as we have + // special knowledge of it anyhow, we don't need to create a special + // conversion object + if ( m_encoding == wxFONTENCODING_ISO8859_1 ) { - if (!m_name) - { -#ifdef __UNIX__ -#if defined(HAVE_LANGINFO_H) && defined(CODESET) - // GNU libc provides current character set this way - char*alang = nl_langinfo(CODESET); - if (alang) - { - SetName(wxConvLibc.cMB2WX(alang)); - } - else -#endif - { - // if we can't get at the character set directly, - // try to see if it's in the environment variables - // (in most cases this won't work, but I was out of ideas) - wxChar *lang = wxGetenv(wxT("LC_ALL")); - if (!lang) lang = wxGetenv(wxT("LC_CTYPE")); - if (!lang) lang = wxGetenv(wxT("LANG")); - wxChar *dot = lang ? wxStrchr(lang, wxT('.')) : (wxChar *)NULL; - if (dot) SetName(dot+1); - } + // don't convert at all + return NULL; + } + + // we trust OS to do conversion better than we can so try external + // conversion methods first + // + // the full order is: + // 1. OS conversion (iconv() under Unix or Win32 API) + // 2. hard coded conversions for UTF + // 3. wxEncodingConverter as fall back + + // step (1) +#ifdef HAVE_ICONV +#if !wxUSE_FONTMAP + if ( m_name ) +#endif // !wxUSE_FONTMAP + { + wxString name(m_name); + +#if wxUSE_FONTMAP + if ( name.empty() ) + name = wxFontMapper::Get()->GetEncodingName(m_encoding); +#endif // wxUSE_FONTMAP + + wxMBConv_iconv *conv = new wxMBConv_iconv(name); + if ( conv->IsOk() ) + return conv; + + delete conv; + } +#endif // HAVE_ICONV + +#ifdef wxHAVE_WIN32_MB2WC + { + wxMBConv_win32 *conv = m_name ? new wxMBConv_win32(m_name) + : new wxMBConv_win32(m_encoding); + if ( conv->IsOk() ) + return conv; + + delete conv; + } +#endif // wxHAVE_WIN32_MB2WC +#if defined(__WXMAC__) + { + if ( m_name || ( m_encoding < wxFONTENCODING_UTF16BE ) ) + { + + wxMBConv_mac *conv = m_name ? new wxMBConv_mac(m_name) + : new wxMBConv_mac(m_encoding); + if ( conv->IsOk() ) + return conv; + + delete conv; + } + } #endif + // step (2) + wxFontEncoding enc = m_encoding; +#if wxUSE_FONTMAP + if ( enc == wxFONTENCODING_SYSTEM && m_name ) + { + // use "false" to suppress interactive dialogs -- we can be called from + // anywhere and popping up a dialog from here is the last thing we want to + // do + enc = wxFontMapper::Get()->CharsetToEncoding(m_name, false); + } +#endif // wxUSE_FONTMAP + + switch ( enc ) + { + case wxFONTENCODING_UTF7: + return new wxMBConvUTF7; + + case wxFONTENCODING_UTF8: + return new wxMBConvUTF8; + + case wxFONTENCODING_UTF16BE: + return new wxMBConvUTF16BE; + + case wxFONTENCODING_UTF16LE: + return new wxMBConvUTF16LE; + + case wxFONTENCODING_UTF32BE: + return new wxMBConvUTF32BE; + + case wxFONTENCODING_UTF32LE: + return new wxMBConvUTF32LE; + + default: + // nothing to do but put here to suppress gcc warnings + ; + } + + // step (3) +#if wxUSE_FONTMAP + { + wxMBConv_wxwin *conv = m_name ? new wxMBConv_wxwin(m_name) + : new wxMBConv_wxwin(m_encoding); + if ( conv->IsOk() ) + return conv; + + delete conv; + } +#endif // wxUSE_FONTMAP + + // NB: This is a hack to prevent deadlock. What could otherwise happen + // in Unicode build: wxConvLocal creation ends up being here + // because of some failure and logs the error. But wxLog will try to + // attach timestamp, for which it will need wxConvLocal (to convert + // time to char* and then wchar_t*), but that fails, tries to log + // error, but wxLog has a (already locked) critical section that + // guards static buffer. + static bool alreadyLoggingError = false; + if (!alreadyLoggingError) + { + alreadyLoggingError = true; + wxLogError(_("Cannot convert from the charset '%s'!"), + m_name ? m_name + : +#if wxUSE_FONTMAP + wxFontMapper::GetEncodingDescription(m_encoding).c_str() +#else // !wxUSE_FONTMAP + wxString::Format(_("encoding %s"), m_encoding).c_str() +#endif // wxUSE_FONTMAP/!wxUSE_FONTMAP + ); + alreadyLoggingError = false; + } + + return NULL; +} + +void wxCSConv::CreateConvIfNeeded() const +{ + if ( m_deferred ) + { + 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()); } - m_cset = wxGetCharacterSet(m_name); - m_deferred = FALSE; +#endif // wxUSE_INTL + + self->m_convReal = DoCreate(); + self->m_deferred = false; } } size_t wxCSConv::MB2WC(wchar_t *buf, const char *psz, size_t n) const { - ((wxCSConv *)this)->LoadNow(); // discard constness - - if (m_cset) - return m_cset->MB2WC(buf, psz, n); + CreateConvIfNeeded(); + + if (m_convReal) + return m_convReal->MB2WC(buf, psz, n); // latin-1 (direct) - size_t len=strlen(psz); - + size_t len = strlen(psz); + if (buf) { - for (size_t c=0; c<=len; c++) + for (size_t c = 0; c <= len; c++) buf[c] = (unsigned char)(psz[c]); } - + return len; } size_t wxCSConv::WC2MB(char *buf, const wchar_t *psz, size_t n) const { - ((wxCSConv *)this)->LoadNow(); // discard constness - - if (m_cset) - return m_cset->WC2MB(buf, psz, n); + CreateConvIfNeeded(); + + if (m_convReal) + return m_convReal->WC2MB(buf, psz, n); // latin-1 (direct) -#if defined(__BORLANDC__) && (__BORLANDC__ > 0x530) - size_t len=std::wcslen(psz); -#else - size_t len=::wcslen(psz); -#endif + const size_t len = wxWcslen(psz); if (buf) { - for (size_t c=0; c<=len; c++) - buf[c] = (psz[c]>0xff) ? '?' : psz[c]; + for (size_t c = 0; c <= len; c++) + { + if (psz[c] > 0xFF) + return (size_t)-1; + buf[c] = psz[c]; + } } - + else + { + for (size_t c = 0; c <= len; c++) + { + if (psz[c] > 0xFF) + return (size_t)-1; + } + } + return len; } -#ifdef HAVE_ICONV_H -class IC_CharSetConverter -{ -public: - IC_CharSetConverter(IC_CharSet*from,IC_CharSet*to) - { cnv = iconv_open(wxConvLibc.cWX2MB(to->cname),wxConvLibc.cWX2MB(from->cname)); } - - ~IC_CharSetConverter() - { if (cnv!=(iconv_t)-1) iconv_close(cnv); } - - size_t Convert(char*buf, const char*psz, size_t n) - { - size_t inbuf = strlen(psz); - size_t outbuf = n; -#ifdef WX_ICONV_TAKES_CHAR - size_t res = iconv( cnv, (char**)&psz, &inbuf, &buf, &outbuf ); +// ---------------------------------------------------------------------------- +// globals +// ---------------------------------------------------------------------------- + +#ifdef __WINDOWS__ + static wxMBConv_win32 wxConvLibcObj; #else - size_t res = iconv( cnv, &psz, &inbuf, &buf, &outbuf ); + static wxMBConvLibc wxConvLibcObj; #endif - if (res==(size_t)-1) return (size_t)-1; - return n-outbuf; - } -public: - iconv_t cnv; -}; -#endif +static wxCSConv wxConvLocalObj(wxFONTENCODING_SYSTEM); +static wxCSConv wxConvISO8859_1Obj(wxFONTENCODING_ISO8859_1); +static wxMBConvUTF7 wxConvUTF7Obj; +static wxMBConvUTF8 wxConvUTF8Obj; -class EC_CharSetConverter -{ -public: - EC_CharSetConverter(EC_CharSet*from,EC_CharSet*to) - { cnv.Init(from->enc,to->enc); } - - size_t Convert(char*buf, const char*psz, size_t n) - { - size_t inbuf = strlen(psz); - if (buf) cnv.Convert(psz,buf); - return inbuf; - } - -public: - wxEncodingConverter cnv; -}; + +WXDLLIMPEXP_DATA_BASE(wxMBConv&) wxConvLibc = wxConvLibcObj; +WXDLLIMPEXP_DATA_BASE(wxCSConv&) wxConvLocal = wxConvLocalObj; +WXDLLIMPEXP_DATA_BASE(wxCSConv&) wxConvISO8859_1 = wxConvISO8859_1Obj; +WXDLLIMPEXP_DATA_BASE(wxMBConvUTF7&) wxConvUTF7 = wxConvUTF7Obj; +WXDLLIMPEXP_DATA_BASE(wxMBConvUTF8&) wxConvUTF8 = wxConvUTF8Obj; +WXDLLIMPEXP_DATA_BASE(wxMBConv *) wxConvCurrent = &wxConvLibcObj; #else // !wxUSE_WCHAR_T -// ---------------------------------------------------------------------------- // stand-ins in absence of wchar_t -// ---------------------------------------------------------------------------- - -WXDLLEXPORT_DATA(wxMBConv) wxConvLibc, wxConvFile; +WXDLLIMPEXP_DATA_BASE(wxMBConv) wxConvLibc, + wxConvISO8859_1, + wxConvLocal, + wxConvUTF8; -#endif // wxUSE_WCHAR_T +#endif // wxUSE_WCHAR_T/!wxUSE_WCHAR_T