X-Git-Url: https://git.saurik.com/wxWidgets.git/blobdiff_plain/69b80d287d7c620b2d39677ed98114dcee3ab1fe..1d49979f7496500d04c39a60fae7c9cc65012dc2:/src/common/strconv.cpp diff --git a/src/common/strconv.cpp b/src/common/strconv.cpp index ffce9eb418..5ceede4225 100644 --- a/src/common/strconv.cpp +++ b/src/common/strconv.cpp @@ -1,11 +1,14 @@ ///////////////////////////////////////////////////////////////////////////// // Name: strconv.cpp // Purpose: Unicode conversion classes -// Author: Ove Kaaven, Robert Roebling, Vadim Zeitlin, Vaclav Slavik +// Author: Ove Kaaven, Robert Roebling, Vadim Zeitlin, Vaclav Slavik, +// Ryan Norton, Fredrik Roubert (UTF7) // Modified by: // Created: 29/01/98 // RCS-ID: $Id$ -// Copyright: (c) 1999 Ove Kaaven, Robert Roebling, Vadim Zeitlin, Vaclav Slavik +// Copyright: (c) 1999 Ove Kaaven, Robert Roebling, Vaclav Slavik +// (c) 2000-2003 Vadim Zeitlin +// (c) 2004 Ryan Norton, Fredrik Roubert // Licence: wxWindows licence ///////////////////////////////////////////////////////////////////////////// @@ -33,10 +36,18 @@ #include "wx/log.h" #endif // WX_PRECOMP +#include "wx/strconv.h" + +#if wxUSE_WCHAR_T + #ifdef __WXMSW__ #include "wx/msw/private.h" #endif +#ifdef __WINDOWS__ + #include "wx/msw/missing.h" +#endif + #ifndef __WXWINCE__ #include #endif @@ -45,64 +56,34 @@ #include #include -#include "wx/module.h" -#include "wx/strconv.h" - -// ---------------------------------------------------------------------------- -// globals -// ---------------------------------------------------------------------------- - -#if wxUSE_WCHAR_T - WXDLLIMPEXP_DATA_BASE(wxMBConv) wxConvLibc; - WXDLLIMPEXP_DATA_BASE(wxCSConv) wxConvLocal((const wxChar *)NULL); - WXDLLIMPEXP_DATA_BASE(wxCSConv) wxConvISO8859_1(_T("iso-8859-1")); -#else - // stand-ins in absence of wchar_t - WXDLLIMPEXP_DATA_BASE(wxMBConv) wxConvLibc, - wxConvFile, - wxConvISO8859_1, - wxConvLocal, - wxConvUTF8; -#endif // wxUSE_WCHAR_T - -WXDLLIMPEXP_DATA_BASE(wxMBConv *) wxConvCurrent = &wxConvLibc; - -class wxStrConvModule: public wxModule -{ -public: - wxStrConvModule() : wxModule() { } - virtual bool OnInit() { return TRUE; } - virtual void OnExit() - { -#if wxUSE_WCHAR_T - wxConvLocal.Clear(); - wxConvISO8859_1.Clear(); -#endif - } - - DECLARE_DYNAMIC_CLASS(wxStrConvModule) -}; - -IMPLEMENT_DYNAMIC_CLASS(wxStrConvModule, wxModule) - +#if defined(__WIN32__) && !defined(__WXMICROWIN__) + #define wxHAVE_WIN32_MB2WC +#endif // __WIN32__ but !__WXMICROWIN__ // ---------------------------------------------------------------------------- // headers // ---------------------------------------------------------------------------- -#if wxUSE_WCHAR_T - #ifdef __SALFORDC__ #include #endif #ifdef HAVE_ICONV #include + #include "wx/thread.h" #endif #include "wx/encconv.h" #include "wx/fontmap.h" +#include "wx/utils.h" +#ifdef __WXMAC__ +#include +#include +#include + +#include "wx/mac/private.h" // includes mac headers +#endif // ---------------------------------------------------------------------------- // macros // ---------------------------------------------------------------------------- @@ -110,12 +91,6 @@ IMPLEMENT_DYNAMIC_CLASS(wxStrConvModule, wxModule) #define BSWAP_UCS4(str, len) { unsigned _c; for (_c=0; _c=0x110000) @@ -163,7 +139,7 @@ static size_t encode_utf16(wxUint32 input, wxUint16 *output) if (output) { *output++ = (wxUint16) ((input >> 10)+0xd7c0); - *output++ = (wxUint16) ((input&0x3ff)+0xdc00); + *output = (wxUint16) ((input&0x3ff)+0xdc00); } return 2; } @@ -193,47 +169,9 @@ static size_t decode_utf16(const wxUint16* input, wxUint32& output) // wxMBConv // ---------------------------------------------------------------------------- -#define IGNORE_LIBC 0 - wxMBConv::~wxMBConv() { - // nothing to do here -} - -size_t wxMBConv::MB2WC(wchar_t *buf, const char *psz, size_t n) const -{ -#if IGNORE_LIBC - if (buf) - { - for (size_t i = 0; i < strlen( psz )+1; i++) - buf[i] = (wchar_t) psz[i]; - return strlen( psz ); - } - else - { - return strlen( psz ); - } -#else - return wxMB2WC(buf, psz, n); -#endif -} - -size_t wxMBConv::WC2MB(char *buf, const wchar_t *psz, size_t n) const -{ -#if IGNORE_LIBC - if (buf) - { - for (size_t i = 0; i < wxStrlen( psz )+1; i++) - buf[i] = (char) psz[i]; - return wxStrlen( psz ); - } - else - { - return wxStrlen( psz ); - } -#else - return wxWC2MB(buf, psz, n); -#endif + // nothing to do here (necessary for Darwin linking probably) } const wxWCharBuffer wxMBConv::cMB2WC(const char *psz) const @@ -246,9 +184,11 @@ const wxWCharBuffer wxMBConv::cMB2WC(const char *psz) const { // now do the actual conversion wxWCharBuffer buf(nLen); - MB2WC(buf.data(), psz, nLen + 1); // with the trailing NUL - - return buf; + nLen = MB2WC(buf.data(), psz, nLen + 1); // with the trailing NULL + if ( nLen != (size_t)-1 ) + { + return buf; + } } } @@ -265,9 +205,11 @@ const wxCharBuffer wxMBConv::cWC2MB(const wchar_t *pwz) const if ( nLen != (size_t)-1 ) { wxCharBuffer buf(nLen+3); // space for a wxUint32 trailing zero - WC2MB(buf.data(), pwz, nLen + 4); - - return buf; + nLen = WC2MB(buf.data(), pwz, nLen + 4); + if ( nLen != (size_t)-1 ) + { + return buf; + } } } @@ -276,43 +218,352 @@ const wxCharBuffer wxMBConv::cWC2MB(const wchar_t *pwz) const return buf; } +const wxWCharBuffer wxMBConv::cMB2WC(const char *szString, size_t nStringLen, size_t* pOutSize) const +{ + wxASSERT(pOutSize != NULL); + + const char* szEnd = szString + nStringLen + 1; + const char* szPos = szString; + const char* szStart = szPos; + + size_t nActualLength = 0; + size_t nCurrentSize = nStringLen; //try normal size first (should never resize?) + + wxWCharBuffer theBuffer(nCurrentSize); + + //Convert the string until the length() is reached, continuing the + //loop every time a null character is reached + while(szPos != szEnd) + { + wxASSERT(szPos < szEnd); //something is _really_ screwed up if this rings true + + //Get the length of the current (sub)string + size_t nLen = MB2WC(NULL, szPos, 0); + + //Invalid conversion? + if( nLen == (size_t)-1 ) + { + *pOutSize = 0; + theBuffer.data()[0u] = wxT('\0'); + return theBuffer; + } + + + //Increase the actual length (+1 for current null character) + nActualLength += nLen + 1; + + //if buffer too big, realloc the buffer + if (nActualLength > (nCurrentSize+1)) + { + wxWCharBuffer theNewBuffer(nCurrentSize << 1); + memcpy(theNewBuffer.data(), theBuffer.data(), nCurrentSize * sizeof(wchar_t)); + theBuffer = theNewBuffer; + nCurrentSize <<= 1; + } + + //Convert the current (sub)string + if ( MB2WC(&theBuffer.data()[szPos - szStart], szPos, nLen + 1) == (size_t)-1 ) + { + *pOutSize = 0; + theBuffer.data()[0u] = wxT('\0'); + return theBuffer; + } + + //Increment to next (sub)string + //Note that we have to use strlen here instead of nLen + //here because XX2XX gives us the size of the output buffer, + //not neccessarly the length of the string + szPos += strlen(szPos) + 1; + } + + //success - return actual length and the buffer + *pOutSize = nActualLength; + return theBuffer; +} + +const wxCharBuffer wxMBConv::cWC2MB(const wchar_t *szString, size_t nStringLen, size_t* pOutSize) const +{ + wxASSERT(pOutSize != NULL); + + const wchar_t* szEnd = szString + nStringLen + 1; + const wchar_t* szPos = szString; + const wchar_t* szStart = szPos; + + size_t nActualLength = 0; + size_t nCurrentSize = nStringLen << 2; //try * 4 first + + wxCharBuffer theBuffer(nCurrentSize); + + //Convert the string until the length() is reached, continuing the + //loop every time a null character is reached + while(szPos != szEnd) + { + wxASSERT(szPos < szEnd); //something is _really_ screwed up if this rings true + + //Get the length of the current (sub)string + size_t nLen = WC2MB(NULL, szPos, 0); + + //Invalid conversion? + if( nLen == (size_t)-1 ) + { + *pOutSize = 0; + theBuffer.data()[0u] = wxT('\0'); + return theBuffer; + } + + //Increase the actual length (+1 for current null character) + nActualLength += nLen + 1; + + //if buffer too big, realloc the buffer + if (nActualLength > (nCurrentSize+1)) + { + wxCharBuffer theNewBuffer(nCurrentSize << 1); + memcpy(theNewBuffer.data(), theBuffer.data(), nCurrentSize); + theBuffer = theNewBuffer; + nCurrentSize <<= 1; + } + + //Convert the current (sub)string + if(WC2MB(&theBuffer.data()[szPos - szStart], szPos, nLen + 1) == (size_t)-1 ) + { + *pOutSize = 0; + theBuffer.data()[0u] = wxT('\0'); + return theBuffer; + } + + //Increment to next (sub)string + //Note that we have to use wxWcslen here instead of nLen + //here because XX2XX gives us the size of the output buffer, + //not neccessarly the length of the string + szPos += wxWcslen(szPos) + 1; + } + + //success - return actual length and the buffer + *pOutSize = nActualLength; + return theBuffer; +} + // ---------------------------------------------------------------------------- -// UTF-7 +// wxMBConvLibc // ---------------------------------------------------------------------------- -WXDLLIMPEXP_DATA_BASE(wxMBConvUTF7) wxConvUTF7; +size_t wxMBConvLibc::MB2WC(wchar_t *buf, const char *psz, size_t n) const +{ + return wxMB2WC(buf, psz, n); +} -#if 0 -static char utf7_setD[]="ABCDEFGHIJKLMNOPQRSTUVWXYZ" - "abcdefghijklmnopqrstuvwxyz" - "0123456789'(),-./:?"; -static char utf7_setO[]="!\"#$%&*;<=>@[]^_`{|}"; -static char utf7_setB[]="ABCDEFGHIJKLMNOPQRSTUVWXYZ" - "abcdefghijklmnopqrstuvwxyz" - "0123456789+/"; -#endif +size_t wxMBConvLibc::WC2MB(char *buf, const wchar_t *psz, size_t n) const +{ + return wxWC2MB(buf, psz, n); +} +// ---------------------------------------------------------------------------- +// UTF-7 +// ---------------------------------------------------------------------------- + +// Implementation (C) 2004 Fredrik Roubert + +// +// BASE64 decoding table +// +static const unsigned char utf7unb64[] = +{ + 0xff, 0xff, 0xff, 0xff, 0xff, 0xff, 0xff, 0xff, + 0xff, 0xff, 0xff, 0xff, 0xff, 0xff, 0xff, 0xff, + 0xff, 0xff, 0xff, 0xff, 0xff, 0xff, 0xff, 0xff, + 0xff, 0xff, 0xff, 0xff, 0xff, 0xff, 0xff, 0xff, + 0xff, 0xff, 0xff, 0xff, 0xff, 0xff, 0xff, 0xff, + 0xff, 0xff, 0xff, 0x3e, 0xff, 0xff, 0xff, 0x3f, + 0x34, 0x35, 0x36, 0x37, 0x38, 0x39, 0x3a, 0x3b, + 0x3c, 0x3d, 0xff, 0xff, 0xff, 0xff, 0xff, 0xff, + 0xff, 0x00, 0x01, 0x02, 0x03, 0x04, 0x05, 0x06, + 0x07, 0x08, 0x09, 0x0a, 0x0b, 0x0c, 0x0d, 0x0e, + 0x0f, 0x10, 0x11, 0x12, 0x13, 0x14, 0x15, 0x16, + 0x17, 0x18, 0x19, 0xff, 0xff, 0xff, 0xff, 0xff, + 0xff, 0x1a, 0x1b, 0x1c, 0x1d, 0x1e, 0x1f, 0x20, + 0x21, 0x22, 0x23, 0x24, 0x25, 0x26, 0x27, 0x28, + 0x29, 0x2a, 0x2b, 0x2c, 0x2d, 0x2e, 0x2f, 0x30, + 0x31, 0x32, 0x33, 0xff, 0xff, 0xff, 0xff, 0xff, + 0xff, 0xff, 0xff, 0xff, 0xff, 0xff, 0xff, 0xff, + 0xff, 0xff, 0xff, 0xff, 0xff, 0xff, 0xff, 0xff, + 0xff, 0xff, 0xff, 0xff, 0xff, 0xff, 0xff, 0xff, + 0xff, 0xff, 0xff, 0xff, 0xff, 0xff, 0xff, 0xff, + 0xff, 0xff, 0xff, 0xff, 0xff, 0xff, 0xff, 0xff, + 0xff, 0xff, 0xff, 0xff, 0xff, 0xff, 0xff, 0xff, + 0xff, 0xff, 0xff, 0xff, 0xff, 0xff, 0xff, 0xff, + 0xff, 0xff, 0xff, 0xff, 0xff, 0xff, 0xff, 0xff, + 0xff, 0xff, 0xff, 0xff, 0xff, 0xff, 0xff, 0xff, + 0xff, 0xff, 0xff, 0xff, 0xff, 0xff, 0xff, 0xff, + 0xff, 0xff, 0xff, 0xff, 0xff, 0xff, 0xff, 0xff, + 0xff, 0xff, 0xff, 0xff, 0xff, 0xff, 0xff, 0xff, + 0xff, 0xff, 0xff, 0xff, 0xff, 0xff, 0xff, 0xff, + 0xff, 0xff, 0xff, 0xff, 0xff, 0xff, 0xff, 0xff, + 0xff, 0xff, 0xff, 0xff, 0xff, 0xff, 0xff, 0xff, + 0xff, 0xff, 0xff, 0xff, 0xff, 0xff, 0xff, 0xff +}; -// TODO: write actual implementations of UTF-7 here -size_t wxMBConvUTF7::MB2WC(wchar_t * WXUNUSED(buf), - const char * WXUNUSED(psz), - size_t WXUNUSED(n)) const +size_t wxMBConvUTF7::MB2WC(wchar_t *buf, const char *psz, size_t n) const { - return 0; + size_t len = 0; + + while (*psz && ((!buf) || (len < n))) + { + unsigned char cc = *psz++; + if (cc != '+') + { + // plain ASCII char + if (buf) + *buf++ = cc; + len++; + } + else if (*psz == '-') + { + // encoded plus sign + if (buf) + *buf++ = cc; + len++; + psz++; + } + else + { + // BASE64 encoded string + bool lsb; + unsigned char c; + unsigned int d, l; + for (lsb = false, d = 0, l = 0; + (cc = utf7unb64[(unsigned char)*psz]) != 0xff; psz++) + { + d <<= 6; + d += cc; + for (l += 6; l >= 8; lsb = !lsb) + { + c = (unsigned char)((d >> (l -= 8)) % 256); + if (lsb) + { + if (buf) + *buf++ |= c; + len ++; + } + else + if (buf) + *buf = (wchar_t)(c << 8); + } + } + if (*psz == '-') + psz++; + } + } + if (buf && (len < n)) + *buf = 0; + return len; } -size_t wxMBConvUTF7::WC2MB(char * WXUNUSED(buf), - const wchar_t * WXUNUSED(psz), - size_t WXUNUSED(n)) const +// +// BASE64 encoding table +// +static const unsigned char utf7enb64[] = +{ + 'A', 'B', 'C', 'D', 'E', 'F', 'G', 'H', + 'I', 'J', 'K', 'L', 'M', 'N', 'O', 'P', + 'Q', 'R', 'S', 'T', 'U', 'V', 'W', 'X', + 'Y', 'Z', 'a', 'b', 'c', 'd', 'e', 'f', + 'g', 'h', 'i', 'j', 'k', 'l', 'm', 'n', + 'o', 'p', 'q', 'r', 's', 't', 'u', 'v', + 'w', 'x', 'y', 'z', '0', '1', '2', '3', + '4', '5', '6', '7', '8', '9', '+', '/' +}; + +// +// UTF-7 encoding table +// +// 0 - Set D (directly encoded characters) +// 1 - Set O (optional direct characters) +// 2 - whitespace characters (optional) +// 3 - special characters +// +static const unsigned char utf7encode[128] = +{ + 3, 3, 3, 3, 3, 3, 3, 3, 3, 2, 2, 3, 3, 2, 3, 3, + 3, 3, 3, 3, 3, 3, 3, 3, 3, 3, 3, 3, 3, 3, 3, 3, + 2, 1, 1, 1, 1, 1, 1, 0, 0, 0, 1, 3, 0, 0, 0, 3, + 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 1, 1, 1, 1, 0, + 1, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, + 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 1, 3, 1, 1, 1, + 1, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, + 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 1, 1, 1, 3, 3 +}; + +size_t wxMBConvUTF7::WC2MB(char *buf, const wchar_t *psz, size_t n) const { - return 0; + + + size_t len = 0; + + while (*psz && ((!buf) || (len < n))) + { + wchar_t cc = *psz++; + if (cc < 0x80 && utf7encode[cc] < 1) + { + // 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; + } +#endif + else + { + if (buf) + *buf++ = '+'; + len++; + if (cc != '+') + { + // BASE64 encode string + unsigned int lsb, d, l; + for (d = 0, l = 0;; psz++) + { + for (lsb = 0; lsb < 2; lsb ++) + { + d <<= 8; + d += lsb ? cc & 0xff : (cc & 0xff00) >> 8; + + for (l += 8; l >= 6; ) + { + l -= 6; + if (buf) + *buf++ = utf7enb64[(d >> l) % 64]; + 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; } // ---------------------------------------------------------------------------- // UTF-8 // ---------------------------------------------------------------------------- -WXDLLIMPEXP_DATA_BASE(wxMBConvUTF8) wxConvUTF8; - static wxUint32 utf8_max[]= { 0x7f, 0x7ff, 0xffff, 0x1fffff, 0x3ffffff, 0x7fffffff, 0xffffffff }; @@ -361,7 +612,8 @@ size_t wxMBConvUTF8::MB2WC(wchar_t *buf, const char *psz, size_t n) const return (size_t)-1; } #ifdef WC_UTF16 - size_t pa = encode_utf16(res, buf); + // cast is ok because wchar_t == wxUuint16 if WC_UTF16 + size_t pa = encode_utf16(res, (wxUint16 *)buf); if (pa == (size_t)-1) return (size_t)-1; if (buf) @@ -388,7 +640,8 @@ size_t wxMBConvUTF8::WC2MB(char *buf, const wchar_t *psz, size_t n) const { wxUint32 cc; #ifdef WC_UTF16 - size_t pa = decode_utf16(psz, cc); + // cast is ok for WC_UTF16 + size_t pa = decode_utf16((const wxUint16 *)psz, cc); psz += (pa == (size_t)-1) ? 1 : pa; #else cc=(*psz++) & 0x7fffffff; @@ -428,24 +681,16 @@ size_t wxMBConvUTF8::WC2MB(char *buf, const wchar_t *psz, size_t n) const // ---------------------------------------------------------------------------- #ifdef WORDS_BIGENDIAN -#define wxMBConvUTF16straight wxMBConvUTF16BE -#define wxMBConvUTF16swap wxMBConvUTF16LE + #define wxMBConvUTF16straight wxMBConvUTF16BE + #define wxMBConvUTF16swap wxMBConvUTF16LE #else -#define wxMBConvUTF16swap wxMBConvUTF16BE -#define wxMBConvUTF16straight wxMBConvUTF16LE + #define wxMBConvUTF16swap wxMBConvUTF16BE + #define wxMBConvUTF16straight wxMBConvUTF16LE #endif -WXDLLIMPEXP_DATA_BASE(wxMBConvUTF16LE) wxConvUTF16LE; -WXDLLIMPEXP_DATA_BASE(wxMBConvUTF16BE) wxConvUTF16BE; - - - - - #ifdef WC_UTF16 - // copy 16bit MB to 16bit String size_t wxMBConvUTF16straight::MB2WC(wchar_t *buf, const char *psz, size_t n) const { @@ -571,7 +816,7 @@ size_t wxMBConvUTF16straight::WC2MB(char *buf, const wchar_t *psz, size_t n) con if (buf) { *(wxUint16*)buf = cc[0]; - buf += sizeof(wxUint16); + buf += sizeof(wxUint16); if (pa > 1) { *(wxUint16*)buf = cc[1]; @@ -707,7 +952,8 @@ size_t wxMBConvUTF32straight::WC2MB(char *buf, const wchar_t *psz, size_t n) con { wxUint32 cc; - size_t pa=decode_utf16(psz, cc); + // cast is ok for WC_UTF16 + size_t pa = decode_utf16((const wxUint16 *)psz, cc); if (pa == (size_t)-1) return pa; @@ -719,7 +965,9 @@ size_t wxMBConvUTF32straight::WC2MB(char *buf, const wchar_t *psz, size_t n) con len += sizeof(wxUint32); psz += pa; } - if (buf && len<=n-sizeof(wxUint32)) *(wxUint32*)buf=0; + + if (buf && len<=n-sizeof(wxUint32)) + *(wxUint32*)buf=0; return len; } @@ -753,7 +1001,9 @@ size_t wxMBConvUTF32swap::MB2WC(wchar_t *buf, const char *psz, size_t n) const len += pa; psz += sizeof(wxUint32); } - 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() const - { return work!=NULL; } -public: - wxMBConv*work; -}; - - // ============================================================================ // The classes doing conversion using the iconv_xxx() functions // ============================================================================ #ifdef HAVE_ICONV -// 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. +// 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) && \ @@ -945,19 +1165,19 @@ public: #define ICONV_CHAR_CAST(x) ((ICONV_CONST char **)(x)) // ---------------------------------------------------------------------------- -// IC_CharSet: encapsulates an iconv character set +// wxMBConv_iconv: encapsulates an iconv character set // ---------------------------------------------------------------------------- -class IC_CharSet : public wxCharacterSet +class wxMBConv_iconv : public wxMBConv { public: - IC_CharSet(const wxChar *name); - virtual ~IC_CharSet(); + wxMBConv_iconv(const wxChar *name); + virtual ~wxMBConv_iconv(); - virtual size_t MB2WC(wchar_t *buf, const char *psz, size_t n); - virtual size_t WC2MB(char *buf, const wchar_t *psz, size_t n); + virtual size_t MB2WC(wchar_t *buf, const char *psz, size_t n) const; + virtual size_t WC2MB(char *buf, const wchar_t *psz, size_t n) const; - bool usable() const + bool IsOk() const { return (m2w != (iconv_t)-1) && (w2m != (iconv_t)-1); } protected: @@ -965,9 +1185,13 @@ protected: // the other direction iconv_t m2w, w2m; +#if wxUSE_THREADS + // guards access to m2w and w2m objects + wxMutex m_iconvMutex; +#endif private: - // the name (for iconv_open()) of a wide char charset - if none is + // the name (for iconv_open()) of a wide char charset -- if none is // available on this machine, it will remain NULL static const char *ms_wcCharsetName; @@ -976,10 +1200,10 @@ private: static bool ms_wcNeedsSwap; }; -const char *IC_CharSet::ms_wcCharsetName = NULL; -bool IC_CharSet::ms_wcNeedsSwap = FALSE; +const char *wxMBConv_iconv::ms_wcCharsetName = NULL; +bool wxMBConv_iconv::ms_wcNeedsSwap = false; -IC_CharSet::IC_CharSet(const wxChar *name) +wxMBConv_iconv::wxMBConv_iconv(const wxChar *name) { // Do it the hard way char cname[100]; @@ -989,7 +1213,7 @@ IC_CharSet::IC_CharSet(const wxChar *name) // check for charset that represents wchar_t: if (ms_wcCharsetName == NULL) { - ms_wcNeedsSwap = FALSE; + ms_wcNeedsSwap = false; // try charset with explicit bytesex info (e.g. "UCS-4LE"): ms_wcCharsetName = WC_NAME_BEST; @@ -1042,7 +1266,7 @@ IC_CharSet::IC_CharSet(const wxChar *name) { ms_wcCharsetName = NULL; - // VS: we must not output an error here, since wxWindows will safely + // VS: we must not output an error here, since wxWidgets will safely // fall back to using wxEncodingConverter. wxLogTrace(wxT("strconv"), wxT("Impossible to convert to/from charset '%s' with iconv, falling back to wxEncodingConverter."), name); //wxLogError( @@ -1066,7 +1290,7 @@ IC_CharSet::IC_CharSet(const wxChar *name) } } -IC_CharSet::~IC_CharSet() +wxMBConv_iconv::~wxMBConv_iconv() { if ( m2w != (iconv_t)-1 ) iconv_close(m2w); @@ -1074,8 +1298,18 @@ IC_CharSet::~IC_CharSet() iconv_close(w2m); } -size_t IC_CharSet::MB2WC(wchar_t *buf, const char *psz, size_t n) +size_t wxMBConv_iconv::MB2WC(wchar_t *buf, const char *psz, size_t n) const { +#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 + // 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 + // as MB<->WC conversion would fail "randomly". + wxMutexLocker lock(wxConstCast(this, wxMBConv_iconv)->m_iconvMutex); +#endif + size_t inbuf = strlen(psz); size_t outbuf = n * SIZEOF_WCHAR_T; size_t res, cres; @@ -1131,8 +1365,13 @@ size_t IC_CharSet::MB2WC(wchar_t *buf, const char *psz, size_t n) return res; } -size_t IC_CharSet::WC2MB(char *buf, const wchar_t *psz, size_t n) +size_t wxMBConv_iconv::WC2MB(char *buf, const wchar_t *psz, size_t n) const { +#if wxUSE_THREADS + // NB: explained in MB2WC + wxMutexLocker lock(wxConstCast(this, wxMBConv_iconv)->m_iconvMutex); +#endif + size_t inbuf = wxWcslen(psz) * SIZEOF_WCHAR_T; size_t outbuf = n; size_t res, cres; @@ -1196,270 +1435,827 @@ size_t IC_CharSet::WC2MB(char *buf, const wchar_t *psz, size_t n) #endif // HAVE_ICONV + // ============================================================================ // Win32 conversion classes // ============================================================================ -#if defined(__WIN32__) && !defined(__WXMICROWIN__) && !defined(__WXUNIVERSAL__) +#ifdef wxHAVE_WIN32_MB2WC // from utils.cpp +#if wxUSE_FONTMAP extern WXDLLIMPEXP_BASE long wxCharsetToCodepage(const wxChar *charset); extern WXDLLIMPEXP_BASE long wxEncodingToCodepage(wxFontEncoding encoding); +#endif -class CP_CharSet : public wxCharacterSet +class wxMBConv_win32 : public wxMBConv { public: - CP_CharSet(const wxChar* name) - { - m_CodePage = wxCharsetToCodepage(name); - } + wxMBConv_win32() + { + m_CodePage = CP_ACP; + } - CP_CharSet(wxFontEncoding encoding) - { - m_CodePage = wxEncodingToCodepage(encoding); - } +#if wxUSE_FONTMAP + wxMBConv_win32(const wxChar* name) + { + m_CodePage = wxCharsetToCodepage(name); + } - size_t MB2WC(wchar_t *buf, const char *psz, size_t n) + wxMBConv_win32(wxFontEncoding encoding) { + m_CodePage = wxEncodingToCodepage(encoding); + } +#endif + + size_t MB2WC(wchar_t *buf, const char *psz, size_t n) const + { + // note that we have to use MB_ERR_INVALID_CHARS flag as it without it + // the behaviour is not compatible with the Unix version (using iconv) + // and break the library itself, e.g. wxTextInputStream::NextChar() + // wouldn't work if reading an incomplete MB char didn't result in an + // error + // + // note however that using MB_ERR_INVALID_CHARS with CP_UTF7 results in + // an error (tested under Windows Server 2003) and apparently it is + // done on purpose, i.e. the function accepts any input in this case + // and although I'd prefer to return error on ill-formed output, our + // own wxMBConvUTF7 doesn't detect errors (e.g. lone "+" which is + // explicitly ill-formed according to RFC 2152) neither so we don't + // even have any fallback here... + int flags = m_CodePage == CP_UTF7 ? 0 : MB_ERR_INVALID_CHARS; + const size_t len = ::MultiByteToWideChar ( m_CodePage, // code page - 0, // flags (none) + flags, // flags: fall on error psz, // input string -1, // its length (NUL-terminated) buf, // output string buf ? n : 0 // size of output buffer ); - // note that it returns # of written chars for buf != NULL and *size* - // of the needed buffer for buf == NULL - return len ? (buf ? len : len - 1) : (size_t)-1; + // note that it returns count of written chars for buf != NULL and size + // of the needed buffer for buf == NULL so in either case the length of + // the string (which never includes the terminating NUL) is one less + return len ? len - 1 : (size_t)-1; } - size_t WC2MB(char *buf, const wchar_t *psz, size_t n) + size_t WC2MB(char *buf, const wchar_t *pwz, size_t n) const { - const size_t len = ::WideCharToMultiByte - ( - m_CodePage, // code page - 0, // flags (none) - psz, // input string + /* + we have a problem here: by default, WideCharToMultiByte() may + replace characters unrepresentable in the target code page with bad + quality approximations such as turning "1/2" symbol (U+00BD) into + "1" for the code pages which don't have it and we, obviously, want + to avoid this at any price + + the trouble is that this function does it _silently_, i.e. it won't + even tell us whether it did or not... Win98/2000 and higher provide + WC_NO_BEST_FIT_CHARS but it doesn't work for the older systems and + we have to resort to a round trip, i.e. check that converting back + results in the same string -- this is, of course, expensive but + otherwise we simply can't be sure to not garble the data. + */ + + // determine if we can rely on WC_NO_BEST_FIT_CHARS: according to MSDN + // it doesn't work with CJK encodings (which we test for rather roughly + // here...) nor with UTF-7/8 nor, of course, with Windows versions not + // supporting it + BOOL usedDef wxDUMMY_INITIALIZE(false); + BOOL *pUsedDef; + int flags; + if ( CanUseNoBestFit() && m_CodePage < 50000 ) + { + // it's our lucky day + flags = WC_NO_BEST_FIT_CHARS; + pUsedDef = &usedDef; + } + else // old system or unsupported encoding + { + flags = 0; + pUsedDef = NULL; + } + + const size_t len = ::WideCharToMultiByte + ( + m_CodePage, // code page + flags, // either none or no best fit + pwz, // input string -1, // it is (wide) NUL-terminated buf, // output buffer buf ? n : 0, // and its size NULL, // default "replacement" char - NULL // [out] was it used? + pUsedDef // [out] was it used? ); - // see the comment above! - return len ? (buf ? len : len - 1) : (size_t)-1; + if ( !len ) + { + // function totally failed + return (size_t)-1; + } + + // if we were really converting, check if we succeeded + if ( buf ) + { + if ( flags ) + { + // check if the conversion failed, i.e. if any replacements + // were done + if ( usedDef ) + return (size_t)-1; + } + else // we must resort to double tripping... + { + wxWCharBuffer wcBuf(n); + if ( MB2WC(wcBuf.data(), buf, n) == (size_t)-1 || + wcscmp(wcBuf, pwz) != 0 ) + { + // we didn't obtain the same thing we started from, hence + // the conversion was lossy and we consider that it failed + return (size_t)-1; + } + } + } + + // see the comment above for the reason of "len - 1" + return len - 1; } - bool usable() const - { return m_CodePage != -1; } + bool IsOk() const { return m_CodePage != -1; } + +private: + static bool CanUseNoBestFit() + { + static int s_isWin98Or2k = -1; + + if ( s_isWin98Or2k == -1 ) + { + int verMaj, verMin; + switch ( wxGetOsVersion(&verMaj, &verMin) ) + { + case wxWIN95: + s_isWin98Or2k = verMaj >= 4 && verMin >= 10; + break; + + case wxWINDOWS_NT: + s_isWin98Or2k = verMaj >= 5; + break; + + default: + // unknown, be conseravtive by default + s_isWin98Or2k = 0; + } + + wxASSERT_MSG( s_isWin98Or2k != -1, _T("should be set above") ); + } + + return s_isWin98Or2k == 1; + } -public: long m_CodePage; }; -#endif // defined(__WIN32__) && !defined(__WXMICROWIN__) && !defined(__WXUNIVERSAL__) + +#endif // wxHAVE_WIN32_MB2WC // ============================================================================ -// wxEncodingConverter based conversion classes +// Cocoa conversion classes // ============================================================================ -#if wxUSE_FONTMAP +#if defined(__WXCOCOA__) + +// RN: There is no UTF-32 support in either Core Foundation or +// Cocoa. Strangely enough, internally Core Foundation uses +// UTF 32 internally quite a bit - its just not public (yet). + +#include +#include -class EC_CharSet : public wxCharacterSet +CFStringEncoding wxCFStringEncFromFontEnc(wxFontEncoding encoding) { -private: - void Init() + CFStringEncoding enc = kCFStringEncodingInvalidId ; + if ( encoding == wxFONTENCODING_DEFAULT ) { - m_ok = m2w.Init(m_enc, wxFONTENCODING_UNICODE) && - w2m.Init(wxFONTENCODING_UNICODE, m_enc); + enc = CFStringGetSystemEncoding(); } + else switch( encoding) + { + case wxFONTENCODING_ISO8859_1 : + enc = kCFStringEncodingISOLatin1 ; + break ; + case wxFONTENCODING_ISO8859_2 : + enc = kCFStringEncodingISOLatin2; + break ; + case wxFONTENCODING_ISO8859_3 : + enc = kCFStringEncodingISOLatin3 ; + break ; + case wxFONTENCODING_ISO8859_4 : + enc = kCFStringEncodingISOLatin4; + break ; + case wxFONTENCODING_ISO8859_5 : + enc = kCFStringEncodingISOLatinCyrillic; + break ; + case wxFONTENCODING_ISO8859_6 : + enc = kCFStringEncodingISOLatinArabic; + break ; + case wxFONTENCODING_ISO8859_7 : + enc = kCFStringEncodingISOLatinGreek; + break ; + case wxFONTENCODING_ISO8859_8 : + enc = kCFStringEncodingISOLatinHebrew; + break ; + case wxFONTENCODING_ISO8859_9 : + enc = kCFStringEncodingISOLatin5; + break ; + case wxFONTENCODING_ISO8859_10 : + enc = kCFStringEncodingISOLatin6; + break ; + case wxFONTENCODING_ISO8859_11 : + enc = kCFStringEncodingISOLatinThai; + break ; + case wxFONTENCODING_ISO8859_13 : + enc = kCFStringEncodingISOLatin7; + break ; + case wxFONTENCODING_ISO8859_14 : + enc = kCFStringEncodingISOLatin8; + break ; + case wxFONTENCODING_ISO8859_15 : + enc = kCFStringEncodingISOLatin9; + break ; + + case wxFONTENCODING_KOI8 : + enc = kCFStringEncodingKOI8_R; + break ; + case wxFONTENCODING_ALTERNATIVE : // MS-DOS CP866 + enc = kCFStringEncodingDOSRussian; + break ; + +// case wxFONTENCODING_BULGARIAN : +// enc = ; +// break ; + + case wxFONTENCODING_CP437 : + enc =kCFStringEncodingDOSLatinUS ; + break ; + case wxFONTENCODING_CP850 : + enc = kCFStringEncodingDOSLatin1; + break ; + case wxFONTENCODING_CP852 : + enc = kCFStringEncodingDOSLatin2; + break ; + case wxFONTENCODING_CP855 : + enc = kCFStringEncodingDOSCyrillic; + break ; + case wxFONTENCODING_CP866 : + enc =kCFStringEncodingDOSRussian ; + break ; + case wxFONTENCODING_CP874 : + enc = kCFStringEncodingDOSThai; + break ; + case wxFONTENCODING_CP932 : + enc = kCFStringEncodingDOSJapanese; + break ; + case wxFONTENCODING_CP936 : + enc =kCFStringEncodingDOSChineseSimplif ; + break ; + case wxFONTENCODING_CP949 : + enc = kCFStringEncodingDOSKorean; + break ; + case wxFONTENCODING_CP950 : + enc = kCFStringEncodingDOSChineseTrad; + break ; + case wxFONTENCODING_CP1250 : + enc = kCFStringEncodingWindowsLatin2; + break ; + case wxFONTENCODING_CP1251 : + enc =kCFStringEncodingWindowsCyrillic ; + break ; + case wxFONTENCODING_CP1252 : + enc =kCFStringEncodingWindowsLatin1 ; + break ; + case wxFONTENCODING_CP1253 : + enc = kCFStringEncodingWindowsGreek; + break ; + case wxFONTENCODING_CP1254 : + enc = kCFStringEncodingWindowsLatin5; + break ; + case wxFONTENCODING_CP1255 : + enc =kCFStringEncodingWindowsHebrew ; + break ; + case wxFONTENCODING_CP1256 : + enc =kCFStringEncodingWindowsArabic ; + break ; + case wxFONTENCODING_CP1257 : + enc = kCFStringEncodingWindowsBalticRim; + break ; +// This only really encodes to UTF7 (if that) evidently +// case wxFONTENCODING_UTF7 : +// enc = kCFStringEncodingNonLossyASCII ; +// break ; + case wxFONTENCODING_UTF8 : + enc = kCFStringEncodingUTF8 ; + break ; + case wxFONTENCODING_EUC_JP : + enc = kCFStringEncodingEUC_JP; + break ; + case wxFONTENCODING_UTF16 : + enc = kCFStringEncodingUnicode ; + break ; + case wxFONTENCODING_MACROMAN : + enc = kCFStringEncodingMacRoman ; + break ; + case wxFONTENCODING_MACJAPANESE : + enc = kCFStringEncodingMacJapanese ; + break ; + case wxFONTENCODING_MACCHINESETRAD : + enc = kCFStringEncodingMacChineseTrad ; + break ; + case wxFONTENCODING_MACKOREAN : + enc = kCFStringEncodingMacKorean ; + break ; + case wxFONTENCODING_MACARABIC : + enc = kCFStringEncodingMacArabic ; + break ; + case wxFONTENCODING_MACHEBREW : + enc = kCFStringEncodingMacHebrew ; + break ; + case wxFONTENCODING_MACGREEK : + enc = kCFStringEncodingMacGreek ; + break ; + case wxFONTENCODING_MACCYRILLIC : + enc = kCFStringEncodingMacCyrillic ; + break ; + case wxFONTENCODING_MACDEVANAGARI : + enc = kCFStringEncodingMacDevanagari ; + break ; + case wxFONTENCODING_MACGURMUKHI : + enc = kCFStringEncodingMacGurmukhi ; + break ; + case wxFONTENCODING_MACGUJARATI : + enc = kCFStringEncodingMacGujarati ; + break ; + case wxFONTENCODING_MACORIYA : + enc = kCFStringEncodingMacOriya ; + break ; + case wxFONTENCODING_MACBENGALI : + enc = kCFStringEncodingMacBengali ; + break ; + case wxFONTENCODING_MACTAMIL : + enc = kCFStringEncodingMacTamil ; + break ; + case wxFONTENCODING_MACTELUGU : + enc = kCFStringEncodingMacTelugu ; + break ; + case wxFONTENCODING_MACKANNADA : + enc = kCFStringEncodingMacKannada ; + break ; + case wxFONTENCODING_MACMALAJALAM : + enc = kCFStringEncodingMacMalayalam ; + break ; + case wxFONTENCODING_MACSINHALESE : + enc = kCFStringEncodingMacSinhalese ; + break ; + case wxFONTENCODING_MACBURMESE : + enc = kCFStringEncodingMacBurmese ; + break ; + case wxFONTENCODING_MACKHMER : + enc = kCFStringEncodingMacKhmer ; + break ; + case wxFONTENCODING_MACTHAI : + enc = kCFStringEncodingMacThai ; + break ; + case wxFONTENCODING_MACLAOTIAN : + enc = kCFStringEncodingMacLaotian ; + break ; + case wxFONTENCODING_MACGEORGIAN : + enc = kCFStringEncodingMacGeorgian ; + break ; + case wxFONTENCODING_MACARMENIAN : + enc = kCFStringEncodingMacArmenian ; + break ; + case wxFONTENCODING_MACCHINESESIMP : + enc = kCFStringEncodingMacChineseSimp ; + break ; + case wxFONTENCODING_MACTIBETAN : + enc = kCFStringEncodingMacTibetan ; + break ; + case wxFONTENCODING_MACMONGOLIAN : + enc = kCFStringEncodingMacMongolian ; + break ; + case wxFONTENCODING_MACETHIOPIC : + enc = kCFStringEncodingMacEthiopic ; + break ; + case wxFONTENCODING_MACCENTRALEUR : + enc = kCFStringEncodingMacCentralEurRoman ; + break ; + case wxFONTENCODING_MACVIATNAMESE : + enc = kCFStringEncodingMacVietnamese ; + break ; + case wxFONTENCODING_MACARABICEXT : + enc = kCFStringEncodingMacExtArabic ; + break ; + case wxFONTENCODING_MACSYMBOL : + enc = kCFStringEncodingMacSymbol ; + break ; + case wxFONTENCODING_MACDINGBATS : + enc = kCFStringEncodingMacDingbats ; + break ; + case wxFONTENCODING_MACTURKISH : + enc = kCFStringEncodingMacTurkish ; + break ; + case wxFONTENCODING_MACCROATIAN : + enc = kCFStringEncodingMacCroatian ; + break ; + case wxFONTENCODING_MACICELANDIC : + enc = kCFStringEncodingMacIcelandic ; + break ; + case wxFONTENCODING_MACROMANIAN : + enc = kCFStringEncodingMacRomanian ; + break ; + case wxFONTENCODING_MACCELTIC : + enc = kCFStringEncodingMacCeltic ; + break ; + case wxFONTENCODING_MACGAELIC : + enc = kCFStringEncodingMacGaelic ; + break ; +// case wxFONTENCODING_MACKEYBOARD : +// enc = kCFStringEncodingMacKeyboardGlyphs ; +// break ; + default : + // because gcc is picky + break ; + } ; + return enc ; +} +class wxMBConv_cocoa : public wxMBConv +{ public: - // temporarily just use wxEncodingConverter stuff, - // so that it works while a better implementation is built - EC_CharSet(const wxChar* name) + wxMBConv_cocoa() { - if (name) - m_enc = wxFontMapper::Get()->CharsetToEncoding(name, FALSE); - else - m_enc = wxFONTENCODING_SYSTEM; + Init(CFStringGetSystemEncoding()) ; + } - Init(); + wxMBConv_cocoa(const wxChar* name) + { + Init( wxCFStringEncFromFontEnc(wxFontMapper::Get()->CharsetToEncoding(name, false) ) ) ; } - EC_CharSet(wxFontEncoding enc) + wxMBConv_cocoa(wxFontEncoding encoding) { - m_enc = enc; + Init( wxCFStringEncFromFontEnc(encoding) ); + } - Init(); + ~wxMBConv_cocoa() + { } - size_t MB2WC(wchar_t *buf, const char *psz, size_t WXUNUSED(n)) + void Init( CFStringEncoding encoding) { - size_t inbuf = strlen(psz); - if (buf) - m2w.Convert(psz,buf); - return inbuf; + m_encoding = encoding ; } - size_t WC2MB(char *buf, const wchar_t *psz, size_t WXUNUSED(n)) + size_t MB2WC(wchar_t * szOut, const char * szUnConv, size_t nOutSize) const { - const size_t inbuf = wxWcslen(psz); - if (buf) - w2m.Convert(psz,buf); + wxASSERT(szUnConv); - return inbuf; + CFStringRef theString = CFStringCreateWithBytes ( + NULL, //the allocator + (const UInt8*)szUnConv, + strlen(szUnConv), + m_encoding, + false //no BOM/external representation + ); + + wxASSERT(theString); + + size_t nOutLength = CFStringGetLength(theString); + + if (szOut == NULL) + { + CFRelease(theString); + return nOutLength; + } + + CFRange theRange = { 0, nOutSize }; + +#if SIZEOF_WCHAR_T == 4 + UniChar* szUniCharBuffer = new UniChar[nOutSize]; +#endif + + CFStringGetCharacters(theString, theRange, szUniCharBuffer); + + CFRelease(theString); + + szUniCharBuffer[nOutLength] = '\0' ; + +#if SIZEOF_WCHAR_T == 4 + wxMBConvUTF16 converter ; + converter.MB2WC(szOut, (const char*)szUniCharBuffer , nOutSize ) ; + delete[] szUniCharBuffer; +#endif + + return nOutLength; } - bool usable() const { return m_ok; } + size_t WC2MB(char *szOut, const wchar_t *szUnConv, size_t nOutSize) const + { + wxASSERT(szUnConv); + + size_t nRealOutSize; + size_t nBufSize = wxWcslen(szUnConv); + UniChar* szUniBuffer = (UniChar*) szUnConv; -public: - wxFontEncoding m_enc; - wxEncodingConverter m2w, w2m; +#if SIZEOF_WCHAR_T == 4 + wxMBConvUTF16BE converter ; + nBufSize = converter.WC2MB( NULL , szUnConv , 0 ); + szUniBuffer = new UniChar[ (nBufSize / sizeof(UniChar)) + 1] ; + converter.WC2MB( (char*) szUniBuffer , szUnConv, nBufSize + sizeof(UniChar)) ; + nBufSize /= sizeof(UniChar); +#endif - // were we initialized successfully? - bool m_ok; + CFStringRef theString = CFStringCreateWithCharactersNoCopy( + NULL, //allocator + szUniBuffer, + nBufSize, + kCFAllocatorNull //deallocator - we want to deallocate it ourselves + ); + + wxASSERT(theString); + + //Note that CER puts a BOM when converting to unicode + //so we check and use getchars instead in that case + if (m_encoding == kCFStringEncodingUnicode) + { + if (szOut != NULL) + CFStringGetCharacters(theString, CFRangeMake(0, nOutSize - 1), (UniChar*) szOut); + + nRealOutSize = CFStringGetLength(theString) + 1; + } + else + { + CFStringGetBytes( + theString, + CFRangeMake(0, CFStringGetLength(theString)), + m_encoding, + 0, //what to put in characters that can't be converted - + //0 tells CFString to return NULL if it meets such a character + false, //not an external representation + (UInt8*) szOut, + nOutSize, + (CFIndex*) &nRealOutSize + ); + } + + CFRelease(theString); + +#if SIZEOF_WCHAR_T == 4 + delete[] szUniBuffer; +#endif + + return nRealOutSize - 1; + } + + bool IsOk() const + { + return m_encoding != kCFStringEncodingInvalidId && + CFStringIsEncodingAvailable(m_encoding); + } - DECLARE_NO_COPY_CLASS(EC_CharSet) +private: + CFStringEncoding m_encoding ; }; -#endif // wxUSE_FONTMAP +#endif // defined(__WXCOCOA__) -// ---------------------------------------------------------------------------- -// the function creating the wxCharacterSet for the specified charset on the -// current system, trying all possibilities -// -// it uses the name if it is given or encoding if name == NULL -// ---------------------------------------------------------------------------- +// ============================================================================ +// Mac conversion classes +// ============================================================================ + +#if defined(__WXMAC__) && defined(TARGET_CARBON) -static wxCharacterSet * -wxGetCharacterSet(const wxChar *name, wxFontEncoding encoding) +class wxMBConv_mac : public wxMBConv { - // check for the special case of ASCII charset - if ( (!name && encoding == wxFONTENCODING_DEFAULT) -#if wxUSE_FONTMAP - || (name && wxFontMapper::Get()-> - CharsetToEncoding(name) == wxFONTENCODING_DEFAULT) -#endif // wxUSE_FONTMAP - ) +public: + wxMBConv_mac() { - // don't convert at all - return NULL; + 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); } - wxCharacterSet *cset = NULL; - if (name) + 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((wxStricmp(name, wxT("UTF8")) == 0) || - (wxStricmp(name, wxT("UTF-8")) == 0) || - encoding == wxFONTENCODING_UTF8 ) + OSStatus status = noErr ; + ByteCount byteOutLen ; + ByteCount byteInLen = strlen(psz) ; + wchar_t *tbuf = NULL ; + UniChar* ubuf = NULL ; + size_t res = 0 ; + + if (buf == NULL) { - cset = new ID_CharSet(&wxConvUTF8); + //apple specs say at least 32 + n = wxMax( 32 , byteInLen ) ; + tbuf = (wchar_t*) malloc( n * SIZEOF_WCHAR_T) ; } - else if((wxStricmp(name, wxT("UTF16")) == 0) || - (wxStricmp(name, wxT("UTF-16")) == 0) || - encoding == wxFONTENCODING_UTF16 ) - { -#ifdef WORDS_BIGENDIAN - cset = new ID_CharSet(&wxConvUTF16BE); + ByteCount byteBufferLen = n * sizeof( UniChar ) ; +#if SIZEOF_WCHAR_T == 4 + ubuf = (UniChar*) malloc( byteBufferLen + 2 ) ; #else - cset = new ID_CharSet(&wxConvUTF16LE); + ubuf = (UniChar*) (buf ? buf : tbuf) ; #endif - } - else if((wxStricmp(name, wxT("UTF16BE")) == 0) || - (wxStricmp(name, wxT("UTF-16BE")) == 0) || - encoding == wxFONTENCODING_UTF16BE ) - { - cset = new ID_CharSet(&wxConvUTF16BE); - } - else if((wxStricmp(name, wxT("UTF16LE")) == 0) || - (wxStricmp(name, wxT("UTF-16LE")) == 0) || - encoding == wxFONTENCODING_UTF16LE ) - { - cset = new ID_CharSet(&wxConvUTF16LE); - } - else if((wxStricmp(name, wxT("UTF32")) == 0) || - (wxStricmp(name, wxT("UTF-32")) == 0) || - (wxStricmp(name, wxT("UCS4")) == 0) || - (wxStricmp(name, wxT("UCS-4")) == 0) || - encoding == wxFONTENCODING_UTF32 ) - { -#ifdef WORDS_BIGENDIAN - cset = new ID_CharSet(&wxConvUTF32BE); + 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 - cset = new ID_CharSet(&wxConvUTF32LE); + res = byteOutLen / sizeof( UniChar ) ; #endif - } - else if((wxStricmp(name, wxT("UTF32BE")) == 0) || - (wxStricmp(name, wxT("UTF-32BE")) == 0) || - (wxStricmp(name, wxT("UCS4BE")) == 0) || - (wxStricmp(name, wxT("UCS-4BE")) == 0) || - encoding == wxFONTENCODING_UTF32BE ) + 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) { - cset = new ID_CharSet(&wxConvUTF32BE); + //apple specs say at least 32 + n = wxMax( 32 , ((byteInLen / SIZEOF_WCHAR_T) * 8) + SIZEOF_WCHAR_T ); + tbuf = (char*) malloc( n ) ; } - else if((wxStricmp(name, wxT("UTF32LE")) == 0) || - (wxStricmp(name, wxT("UTF-32LE")) == 0) || - (wxStricmp(name, wxT("UCS4LE")) == 0) || - (wxStricmp(name, wxT("UCS-4LE")) == 0) || - encoding == wxFONTENCODING_UTF32 ) + + ByteCount byteBufferLen = n ; + UniChar* ubuf = NULL ; +#if SIZEOF_WCHAR_T == 4 + wxMBConvUTF16BE converter ; + size_t unicharlen = converter.WC2MB( NULL , psz , 0 ) ; + byteInLen = unicharlen ; + ubuf = (UniChar*) malloc( byteInLen + 2 ) ; + converter.WC2MB( (char*) ubuf , psz, unicharlen + 2 ) ; +#else + ubuf = (UniChar*) psz ; +#endif + status = TECConvertText(m_WC2MB_converter, (ConstTextPtr) ubuf , byteInLen, &byteInLen, + (TextPtr) (buf ? buf : tbuf) , byteBufferLen, &byteOutLen); +#if SIZEOF_WCHAR_T == 4 + free( ubuf ) ; +#endif + if ( buf == NULL ) + free(tbuf) ; + + size_t res = byteOutLen ; + if ( buf && res < n) { - cset = new ID_CharSet(&wxConvUTF32LE); + buf[res] = 0; + + //we need to double-trip to verify it didn't insert any ? in place + //of bogus characters + wxWCharBuffer wcBuf(n); + size_t pszlen = wxWcslen(psz); + if ( MB2WC(wcBuf.data(), buf, n) == (size_t)-1 || + 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; + } } -#ifdef HAVE_ICONV + + 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) { - cset = new IC_CharSet(name); + if (!m2w.Convert(psz,buf)) + return (size_t)-1; } -#endif // HAVE_ICONV + return inbuf; } - // it can only be NULL in this case -#ifndef HAVE_ICONV - if ( cset ) -#endif // !HAVE_ICONV + size_t WC2MB(char *buf, const wchar_t *psz, size_t WXUNUSED(n)) const { - if ( cset->usable() ) - return cset; + const size_t inbuf = wxWcslen(psz); + if (buf) + { + if (!w2m.Convert(psz,buf)) + return (size_t)-1; + } - delete cset; - cset = NULL; + return inbuf; } -#if defined(__WIN32__) && !defined(__WXMICROWIN__) && !defined(__WXUNIVERSAL__) - cset = name ? new CP_CharSet(name) : new CP_CharSet(encoding); - if ( cset->usable() ) - return cset; + bool IsOk() const { return m_ok; } - delete cset; - cset = NULL; -#endif // defined(__WIN32__) && !defined(__WXMICROWIN__) && !defined(__WXUNIVERSAL__) - -#if wxUSE_FONTMAP - cset = name ? new EC_CharSet(name) : new EC_CharSet(encoding); - if ( cset->usable() ) - return cset; +public: + wxFontEncoding m_enc; + wxEncodingConverter m2w, w2m; - delete cset; - cset = NULL; -#endif // wxUSE_FONTMAP + // were we initialized successfully? + bool m_ok; - wxLogError(_("Cannot convert from encoding '%s'!"), - name ? name - : -#if wxUSE_FONTMAP - wxFontMapper::GetEncodingDescription(encoding).c_str() -#else // !wxUSE_FONTMAP - wxString::Format(_T("%s"), encoding).c_str() -#endif // wxUSE_FONTMAP/!wxUSE_FONTMAP - ); + DECLARE_NO_COPY_CLASS(wxMBConv_wxwin) +}; - return NULL; -} +#endif // wxUSE_FONTMAP // ============================================================================ // wxCSConv implementation @@ -1467,21 +2263,32 @@ wxGetCharacterSet(const wxChar *name, wxFontEncoding encoding) void wxCSConv::Init() { - m_name = (wxChar *)NULL; - m_cset = (wxCharacterSet *) NULL; - m_deferred = TRUE; + m_name = NULL; + m_convReal = NULL; + m_deferred = true; } wxCSConv::wxCSConv(const wxChar *charset) { Init(); - m_encoding = wxFONTENCODING_DEFAULT; - SetName(charset); + if ( charset ) + { + SetName(charset); + } + + m_encoding = wxFONTENCODING_SYSTEM; } wxCSConv::wxCSConv(wxFontEncoding encoding) { + if ( encoding == wxFONTENCODING_MAX || encoding == wxFONTENCODING_DEFAULT ) + { + wxFAIL_MSG( _T("invalid encoding value in wxCSConv ctor") ); + + encoding = wxFONTENCODING_SYSTEM; + } + Init(); m_encoding = encoding; @@ -1514,10 +2321,10 @@ wxCSConv& wxCSConv::operator=(const wxCSConv& conv) void wxCSConv::Clear() { free(m_name); - delete m_cset; + delete m_convReal; m_name = NULL; - m_cset = NULL; + m_convReal = NULL; } void wxCSConv::SetName(const wxChar *charset) @@ -1525,39 +2332,193 @@ void wxCSConv::SetName(const wxChar *charset) if (charset) { m_name = wxStrdup(charset); - m_deferred = TRUE; + m_deferred = true; + } +} + +wxMBConv *wxCSConv::DoCreate() const +{ + // check for the special case of ASCII or ISO8859-1 charset: as we have + // special knowledge of it anyhow, we don't need to create a special + // conversion object + if ( m_encoding == wxFONTENCODING_ISO8859_1 ) + { + // don't convert at all + return NULL; + } + + // we trust OS to do conversion better than we can so try external + // conversion methods first + // + // the full order is: + // 1. OS conversion (iconv() under Unix or Win32 API) + // 2. hard coded conversions for UTF + // 3. wxEncodingConverter as fall back + + // step (1) +#ifdef HAVE_ICONV +#if !wxUSE_FONTMAP + if ( m_name ) +#endif // !wxUSE_FONTMAP + { + wxString name(m_name); + +#if wxUSE_FONTMAP + if ( name.empty() ) + name = wxFontMapper::Get()->GetEncodingName(m_encoding); +#endif // wxUSE_FONTMAP + + wxMBConv_iconv *conv = new wxMBConv_iconv(name); + if ( conv->IsOk() ) + return conv; + + delete conv; + } +#endif // HAVE_ICONV + +#ifdef wxHAVE_WIN32_MB2WC + { +#if wxUSE_FONTMAP + wxMBConv_win32 *conv = m_name ? new wxMBConv_win32(m_name) + : new wxMBConv_win32(m_encoding); + if ( conv->IsOk() ) + return conv; + + delete conv; +#else + return NULL; +#endif + } +#endif // wxHAVE_WIN32_MB2WC +#if defined(__WXMAC__) + { + if ( m_name || ( m_encoding < wxFONTENCODING_UTF16BE ) ) + { + + wxMBConv_mac *conv = m_name ? new wxMBConv_mac(m_name) + : new wxMBConv_mac(m_encoding); + if ( conv->IsOk() ) + return conv; + + delete conv; + } + } +#endif +#if defined(__WXCOCOA__) + { + if ( m_name || ( m_encoding <= wxFONTENCODING_UTF16 ) ) + { + + wxMBConv_cocoa *conv = m_name ? new wxMBConv_cocoa(m_name) + : new wxMBConv_cocoa(m_encoding); + if ( conv->IsOk() ) + return conv; + + delete conv; + } + } +#endif + // step (2) + wxFontEncoding enc = m_encoding; +#if wxUSE_FONTMAP + if ( enc == wxFONTENCODING_SYSTEM && m_name ) + { + // use "false" to suppress interactive dialogs -- we can be called from + // anywhere and popping up a dialog from here is the last thing we want to + // do + enc = wxFontMapper::Get()->CharsetToEncoding(m_name, false); + } +#endif // wxUSE_FONTMAP + + switch ( enc ) + { + case wxFONTENCODING_UTF7: + return new wxMBConvUTF7; + + case wxFONTENCODING_UTF8: + return new wxMBConvUTF8; + + case wxFONTENCODING_UTF16BE: + return new wxMBConvUTF16BE; + + case wxFONTENCODING_UTF16LE: + return new wxMBConvUTF16LE; + + case wxFONTENCODING_UTF32BE: + return new wxMBConvUTF32BE; + + case wxFONTENCODING_UTF32LE: + return new wxMBConvUTF32LE; + + default: + // nothing to do but put here to suppress gcc warnings + ; + } + + // step (3) +#if wxUSE_FONTMAP + { + wxMBConv_wxwin *conv = m_name ? new wxMBConv_wxwin(m_name) + : new wxMBConv_wxwin(m_encoding); + if ( conv->IsOk() ) + return conv; + + delete conv; } +#endif // wxUSE_FONTMAP + + // NB: This is a hack to prevent deadlock. What could otherwise happen + // in Unicode build: wxConvLocal creation ends up being here + // because of some failure and logs the error. But wxLog will try to + // attach timestamp, for which it will need wxConvLocal (to convert + // time to char* and then wchar_t*), but that fails, tries to log + // error, but wxLog has a (already locked) critical section that + // guards static buffer. + static bool alreadyLoggingError = false; + if (!alreadyLoggingError) + { + alreadyLoggingError = true; + wxLogError(_("Cannot convert from the charset '%s'!"), + m_name ? m_name + : +#if wxUSE_FONTMAP + wxFontMapper::GetEncodingDescription(m_encoding).c_str() +#else // !wxUSE_FONTMAP + wxString::Format(_("encoding %s"), m_encoding).c_str() +#endif // wxUSE_FONTMAP/!wxUSE_FONTMAP + ); + alreadyLoggingError = false; + } + + return NULL; } -void wxCSConv::LoadNow() +void wxCSConv::CreateConvIfNeeded() const { if ( m_deferred ) { - // it would probably be better to make GetSystemEncodingName() always - // available (i.e. even when wxUSE_INTL == 0)? + wxCSConv *self = (wxCSConv *)this; // const_cast + #if wxUSE_INTL - if ( !m_name && m_encoding == wxFONTENCODING_DEFAULT ) + // if we don't have neither the name nor the encoding, use the default + // encoding for this system + if ( !m_name && m_encoding == wxFONTENCODING_SYSTEM ) { - wxString name = wxLocale::GetSystemEncodingName(); - if ( !name.empty() ) - { - SetName(name); - } + self->m_name = wxStrdup(wxLocale::GetSystemEncodingName()); } #endif // wxUSE_INTL - // wxGetCharacterSet() complains about NULL name - m_cset = wxGetCharacterSet(m_name, m_encoding); - m_deferred = FALSE; + self->m_convReal = DoCreate(); + self->m_deferred = false; } } size_t wxCSConv::MB2WC(wchar_t *buf, const char *psz, size_t n) const { - ((wxCSConv *)this)->LoadNow(); // discard constness + CreateConvIfNeeded(); - if (m_cset) - return m_cset->MB2WC(buf, psz, n); + if (m_convReal) + return m_convReal->MB2WC(buf, psz, n); // latin-1 (direct) size_t len = strlen(psz); @@ -1573,22 +2534,67 @@ size_t wxCSConv::MB2WC(wchar_t *buf, const char *psz, size_t n) const size_t wxCSConv::WC2MB(char *buf, const wchar_t *psz, size_t n) const { - ((wxCSConv *)this)->LoadNow(); // discard constness + CreateConvIfNeeded(); - if (m_cset) - return m_cset->WC2MB(buf, psz, n); + if (m_convReal) + return m_convReal->WC2MB(buf, psz, n); // latin-1 (direct) const size_t len = wxWcslen(psz); if (buf) { for (size_t c = 0; c <= len; c++) - buf[c] = (psz[c] > 0xff) ? '?' : psz[c]; + { + if (psz[c] > 0xFF) + return (size_t)-1; + buf[c] = (char)psz[c]; + } + } + else + { + for (size_t c = 0; c <= len; c++) + { + if (psz[c] > 0xFF) + return (size_t)-1; + } } return len; } -#endif // wxUSE_WCHAR_T +// ---------------------------------------------------------------------------- +// globals +// ---------------------------------------------------------------------------- + +#ifdef __WINDOWS__ + static wxMBConv_win32 wxConvLibcObj; +#elif defined(__WXMAC__) && !defined(__MACH__) + static wxMBConv_mac wxConvLibcObj ; +#else + static wxMBConvLibc wxConvLibcObj; +#endif + +static wxCSConv wxConvLocalObj(wxFONTENCODING_SYSTEM); +static wxCSConv wxConvISO8859_1Obj(wxFONTENCODING_ISO8859_1); +static wxMBConvUTF7 wxConvUTF7Obj; +static wxMBConvUTF8 wxConvUTF8Obj; + + +WXDLLIMPEXP_DATA_BASE(wxMBConv&) wxConvLibc = wxConvLibcObj; +WXDLLIMPEXP_DATA_BASE(wxCSConv&) wxConvLocal = wxConvLocalObj; +WXDLLIMPEXP_DATA_BASE(wxCSConv&) wxConvISO8859_1 = wxConvISO8859_1Obj; +WXDLLIMPEXP_DATA_BASE(wxMBConvUTF7&) wxConvUTF7 = wxConvUTF7Obj; +WXDLLIMPEXP_DATA_BASE(wxMBConvUTF8&) wxConvUTF8 = wxConvUTF8Obj; +WXDLLIMPEXP_DATA_BASE(wxMBConv *) wxConvCurrent = &wxConvLibcObj; + +#else // !wxUSE_WCHAR_T + +// stand-ins in absence of wchar_t +WXDLLIMPEXP_DATA_BASE(wxMBConv) wxConvLibc, + wxConvISO8859_1, + wxConvLocal, + wxConvUTF8; + +#endif // wxUSE_WCHAR_T/!wxUSE_WCHAR_T