X-Git-Url: https://git.saurik.com/wxWidgets.git/blobdiff_plain/c547282d5e0205b5dab200d36b7f76f629f7a8fe..86948c99a6f0fd177b09b3db0ef702739ec62a27:/src/common/strconv.cpp diff --git a/src/common/strconv.cpp b/src/common/strconv.cpp index ffd70ff003..7b0e78a69d 100644 --- a/src/common/strconv.cpp +++ b/src/common/strconv.cpp @@ -1,12 +1,14 @@ ///////////////////////////////////////////////////////////////////////////// -// Name: strconv.cpp +// Name: src/common/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, Vaclav Slavik // (c) 2000-2003 Vadim Zeitlin +// (c) 2004 Ryan Norton, Fredrik Roubert // Licence: wxWindows licence ///////////////////////////////////////////////////////////////////////////// @@ -18,10 +20,6 @@ // headers // ---------------------------------------------------------------------------- -#if defined(__GNUG__) && !defined(NO_GCC_PRAGMA) - #pragma implementation "strconv.h" -#endif - // For compilers that support precompilation, includes "wx.h". #include "wx/wxprec.h" @@ -38,8 +36,9 @@ #if wxUSE_WCHAR_T -#ifdef __WXMSW__ +#ifdef __WINDOWS__ #include "wx/msw/private.h" + #include "wx/msw/missing.h" #endif #ifndef __WXWINCE__ @@ -54,48 +53,33 @@ #define wxHAVE_WIN32_MB2WC #endif // __WIN32__ but !__WXMICROWIN__ -// ---------------------------------------------------------------------------- -// headers -// ---------------------------------------------------------------------------- - #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" -// ---------------------------------------------------------------------------- -// macros -// ---------------------------------------------------------------------------- +#ifdef __WXMAC__ +#ifndef __DARWIN__ +#include +#include +#include +#endif -#define BSWAP_UCS4(str, len) { unsigned _c; for (_c=0; _c=0xdfff)) + else if ((input[1]<0xdc00) || (input[1]>0xdfff)) { output = *input; return (size_t)-1; @@ -156,7 +140,7 @@ static size_t decode_utf16(const wxUint16* input, wxUint32& output) wxMBConv::~wxMBConv() { - // nothing to do here + // nothing to do here (necessary for Darwin linking probably) } const wxWCharBuffer wxMBConv::cMB2WC(const char *psz) const @@ -169,9 +153,11 @@ 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; + } } } @@ -188,9 +174,11 @@ 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; + } } } @@ -199,6 +187,131 @@ 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 instead of nLen here + //because XX2XX gives us the size of the output buffer, + //which is not necessarily 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 instead of nLen here + //because XX2XX gives us the size of the output buffer, + //which is not necessarily the length of the string + szPos += wxWcslen(szPos) + 1; + } + + //success - return actual length and the buffer + *pOutSize = nActualLength; + return theBuffer; +} + // ---------------------------------------------------------------------------- // wxMBConvLibc // ---------------------------------------------------------------------------- @@ -213,33 +326,241 @@ size_t wxMBConvLibc::WC2MB(char *buf, const wchar_t *psz, size_t n) const return wxWC2MB(buf, psz, n); } +#ifdef __UNIX__ + // ---------------------------------------------------------------------------- -// UTF-7 +// wxConvBrokenFileNames // ---------------------------------------------------------------------------- -#if 0 -static char utf7_setD[]="ABCDEFGHIJKLMNOPQRSTUVWXYZ" - "abcdefghijklmnopqrstuvwxyz" - "0123456789'(),-./:?"; -static char utf7_setO[]="!\"#$%&*;<=>@[]^_`{|}"; -static char utf7_setB[]="ABCDEFGHIJKLMNOPQRSTUVWXYZ" - "abcdefghijklmnopqrstuvwxyz" - "0123456789+/"; +wxConvBrokenFileNames::wxConvBrokenFileNames(const wxChar *charset) +{ + if ( !charset || wxStricmp(charset, _T("UTF-8")) == 0 + || wxStricmp(charset, _T("UTF8")) == 0 ) + m_conv = new wxMBConvUTF8(wxMBConvUTF8::MAP_INVALID_UTF8_TO_OCTAL); + else + m_conv = new wxCSConv(charset); +} + +size_t +wxConvBrokenFileNames::MB2WC(wchar_t *outputBuf, + const char *psz, + size_t outputSize) const +{ + return m_conv->MB2WC( outputBuf, psz, outputSize ); +} + +size_t +wxConvBrokenFileNames::WC2MB(char *outputBuf, + const wchar_t *psz, + size_t outputSize) const +{ + return m_conv->WC2MB( outputBuf, psz, outputSize ); +} + #endif -// 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 +// ---------------------------------------------------------------------------- +// 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 +}; + +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; /*nothing*/; 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; } // ---------------------------------------------------------------------------- @@ -249,12 +570,19 @@ size_t wxMBConvUTF7::WC2MB(char * WXUNUSED(buf), static wxUint32 utf8_max[]= { 0x7f, 0x7ff, 0xffff, 0x1fffff, 0x3ffffff, 0x7fffffff, 0xffffffff }; +// boundaries of the private use area we use to (temporarily) remap invalid +// characters invalid in a UTF-8 encoded string +const wxUint32 wxUnicodePUA = 0x100000; +const wxUint32 wxUnicodePUAEnd = wxUnicodePUA + 256; + size_t wxMBConvUTF8::MB2WC(wchar_t *buf, const char *psz, size_t n) const { size_t len = 0; while (*psz && ((!buf) || (len < n))) { + const char *opsz = psz; + bool invalid = false; unsigned char cc = *psz++, fc = cc; unsigned cnt; for (cnt = 0; fc & 0x80; cnt++) @@ -265,6 +593,15 @@ size_t wxMBConvUTF8::MB2WC(wchar_t *buf, const char *psz, size_t n) const if (buf) *buf++ = cc; len++; + + // escape the escape character for octal escapes + if ((m_options & MAP_INVALID_UTF8_TO_OCTAL) + && cc == '\\' && (!buf || len < n)) + { + if (buf) + *buf++ = cc; + len++; + } } else { @@ -272,7 +609,7 @@ size_t wxMBConvUTF8::MB2WC(wchar_t *buf, const char *psz, size_t n) const if (!cnt) { // invalid UTF-8 sequence - return (size_t)-1; + invalid = true; } else { @@ -280,32 +617,93 @@ size_t wxMBConvUTF8::MB2WC(wchar_t *buf, const char *psz, size_t n) const wxUint32 res = cc & (0x3f >> cnt); while (cnt--) { - cc = *psz++; + cc = *psz; if ((cc & 0xC0) != 0x80) { // invalid UTF-8 sequence - return (size_t)-1; + invalid = true; + break; } + psz++; res = (res << 6) | (cc & 0x3f); } - if (res <= utf8_max[ocnt]) + if (invalid || res <= utf8_max[ocnt]) { // illegal UTF-8 encoding - return (size_t)-1; + invalid = true; + } + else if ((m_options & MAP_INVALID_UTF8_TO_PUA) && + res >= wxUnicodePUA && res < wxUnicodePUAEnd) + { + // if one of our PUA characters turns up externally + // it must also be treated as an illegal sequence + // (a bit like you have to escape an escape character) + invalid = true; } + else + { #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) - return (size_t)-1; - if (buf) - buf += pa; - len += pa; + // cast is ok because wchar_t == wxUuint16 if WC_UTF16 + size_t pa = encode_utf16(res, (wxUint16 *)buf); + if (pa == (size_t)-1) + { + invalid = true; + } + else + { + if (buf) + buf += pa; + len += pa; + } #else // !WC_UTF16 - if (buf) - *buf++ = res; - len++; + if (buf) + *buf++ = (wchar_t)res; + len++; #endif // WC_UTF16/!WC_UTF16 + } + } + if (invalid) + { + if (m_options & MAP_INVALID_UTF8_TO_PUA) + { + while (opsz < psz && (!buf || len < n)) + { +#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); + if (buf) + buf += pa; + opsz++; + len += pa; +#else + if (buf) + *buf++ = (wchar_t)(wxUnicodePUA + (unsigned char)*opsz); + opsz++; + len++; +#endif + } + } + else if (m_options & MAP_INVALID_UTF8_TO_OCTAL) + { + while (opsz < psz && (!buf || len < n)) + { + if ( buf && len + 3 < n ) + { + unsigned char on = *opsz; + *buf++ = L'\\'; + *buf++ = (wchar_t)( L'0' + on / 0100 ); + *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; + } } } } @@ -314,6 +712,11 @@ size_t wxMBConvUTF8::MB2WC(wchar_t *buf, const char *psz, size_t n) const return len; } +static inline bool isoctal(wchar_t wch) +{ + return L'0' <= wch && wch <= L'7'; +} + size_t wxMBConvUTF8::WC2MB(char *buf, const wchar_t *psz, size_t n) const { size_t len = 0; @@ -328,36 +731,67 @@ size_t wxMBConvUTF8::WC2MB(char *buf, const wchar_t *psz, size_t n) const #else cc=(*psz++) & 0x7fffffff; #endif - unsigned cnt; - for (cnt = 0; cc > utf8_max[cnt]; cnt++) {} - if (!cnt) + + if ( (m_options & MAP_INVALID_UTF8_TO_PUA) + && cc >= wxUnicodePUA && cc < wxUnicodePUAEnd ) + { + if (buf) + *buf++ = (char)(cc - wxUnicodePUA); + len++; + } + else if ( (m_options & MAP_INVALID_UTF8_TO_OCTAL) + && cc == L'\\' && psz[0] == L'\\' ) { - // plain ASCII char if (buf) - *buf++ = (char) cc; + *buf++ = (char)cc; + psz++; len++; } + else if ( (m_options & MAP_INVALID_UTF8_TO_OCTAL) && + cc == L'\\' && + isoctal(psz[0]) && isoctal(psz[1]) && isoctal(psz[2]) ) + { + if (buf) + { + *buf++ = (char) ((psz[0] - L'0')*0100 + + (psz[1] - L'0')*010 + + (psz[2] - L'0')); + } + psz += 3; + len++; + } else { - len += cnt + 1; - if (buf) + unsigned cnt; + for (cnt = 0; cc > utf8_max[cnt]; cnt++) {} + if (!cnt) { - *buf++ = (char) ((-128 >> cnt) | ((cc >> (cnt * 6)) & (0x3f >> cnt))); - while (cnt--) - *buf++ = (char) (0x80 | ((cc >> (cnt * 6)) & 0x3f)); + // 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 && (lenIsOk() ) + { + delete result; + return 0; + } + return result; +} + +wxString wxMBConv_iconv::ms_wcCharsetName; bool wxMBConv_iconv::ms_wcNeedsSwap = false; wxMBConv_iconv::wxMBConv_iconv(const wxChar *name) { - // Do it the hard way - char cname[100]; - for (size_t i = 0; i < wxStrlen(name)+1; i++) - cname[i] = (char) name[i]; + // iconv operates with chars, not wxChars, but luckily it uses only ASCII + // names for the charsets + const wxCharBuffer cname(wxString(name).ToAscii()); // check for charset that represents wchar_t: - if (ms_wcCharsetName == NULL) + if ( ms_wcCharsetName.empty() ) { - ms_wcNeedsSwap = false; + wxLogTrace(TRACE_STRCONV, _T("Looking for wide char codeset:")); - // try charset with explicit bytesex info (e.g. "UCS-4LE"): - ms_wcCharsetName = WC_NAME_BEST; - m2w = iconv_open(ms_wcCharsetName, cname); +#if wxUSE_FONTMAP + const wxChar **names = wxFontMapperBase::GetAllEncodingNames(WC_ENC); +#else // !wxUSE_FONTMAP + static const wxChar *names[] = + { +#if SIZEOF_WCHAR_T == 4 + _T("UCS-4"), +#elif SIZEOF_WCHAR_T = 2 + _T("UCS-2"), +#endif + NULL + }; +#endif // wxUSE_FONTMAP/!wxUSE_FONTMAP - if (m2w == (iconv_t)-1) + for ( ; *names && ms_wcCharsetName.empty(); ++names ) { - // try charset w/o bytesex info (e.g. "UCS4") - // and check for bytesex ourselves: - ms_wcCharsetName = WC_NAME; - m2w = iconv_open(ms_wcCharsetName, cname); + const wxString nameCS(*names); - // last bet, try if it knows WCHAR_T pseudo-charset - if (m2w == (iconv_t)-1) - { - ms_wcCharsetName = "WCHAR_T"; - m2w = iconv_open(ms_wcCharsetName, cname); - } + // first try charset with explicit bytesex info (e.g. "UCS-4LE"): + wxString nameXE(nameCS); + #ifdef WORDS_BIGENDIAN + nameXE += _T("BE"); + #else // little endian + nameXE += _T("LE"); + #endif - if (m2w != (iconv_t)-1) + wxLogTrace(TRACE_STRCONV, _T(" trying charset \"%s\""), + nameXE.c_str()); + + m2w = iconv_open(nameXE.ToAscii(), cname); + if ( m2w == ICONV_T_INVALID ) { - char buf[2], *bufPtr; - wchar_t wbuf[2], *wbufPtr; - size_t insz, outsz; - size_t res; - - buf[0] = 'A'; - buf[1] = 0; - wbuf[0] = 0; - insz = 2; - outsz = SIZEOF_WCHAR_T * 2; - wbufPtr = wbuf; - bufPtr = buf; - - res = iconv(m2w, ICONV_CHAR_CAST(&bufPtr), &insz, - (char**)&wbufPtr, &outsz); - - if (ICONV_FAILED(res, insz)) - { - ms_wcCharsetName = NULL; - wxLogLastError(wxT("iconv")); - wxLogError(_("Conversion to charset '%s' doesn't work."), name); - } - else + // try charset w/o bytesex info (e.g. "UCS4") + wxLogTrace(TRACE_STRCONV, _T(" trying charset \"%s\""), + nameCS.c_str()); + m2w = iconv_open(nameCS.ToAscii(), cname); + + // and check for bytesex ourselves: + if ( m2w != ICONV_T_INVALID ) { - ms_wcNeedsSwap = wbuf[0] != (wchar_t)buf[0]; + char buf[2], *bufPtr; + wchar_t wbuf[2], *wbufPtr; + size_t insz, outsz; + size_t res; + + buf[0] = 'A'; + buf[1] = 0; + wbuf[0] = 0; + insz = 2; + outsz = SIZEOF_WCHAR_T * 2; + wbufPtr = wbuf; + bufPtr = buf; + + res = iconv(m2w, ICONV_CHAR_CAST(&bufPtr), &insz, + (char**)&wbufPtr, &outsz); + + if (ICONV_FAILED(res, insz)) + { + wxLogLastError(wxT("iconv")); + wxLogError(_("Conversion to charset '%s' doesn't work."), + nameCS.c_str()); + } + else // ok, can convert to this encoding, remember it + { + ms_wcCharsetName = nameCS; + ms_wcNeedsSwap = wbuf[0] != (wchar_t)buf[0]; + } } } - else + else // use charset not requiring byte swapping { - ms_wcCharsetName = NULL; - - // VS: we must not output an error here, since wxWindows 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( + ms_wcCharsetName = nameXE; } } - wxLogTrace(wxT("strconv"), wxT("wchar_t charset is '%s', needs swap: %i"), ms_wcCharsetName, ms_wcNeedsSwap); + + wxLogTrace(TRACE_STRCONV, + wxT("iconv wchar_t charset is \"%s\"%s"), + ms_wcCharsetName.empty() ? _T("") + : ms_wcCharsetName.c_str(), + ms_wcNeedsSwap ? _T(" (needs swap)") + : _T("")); } else // we already have ms_wcCharsetName { - m2w = iconv_open(ms_wcCharsetName, cname); + m2w = iconv_open(ms_wcCharsetName.ToAscii(), cname); } - // NB: don't ever pass NULL to iconv_open(), it may crash! - if ( ms_wcCharsetName ) + if ( ms_wcCharsetName.empty() ) { - w2m = iconv_open( cname, ms_wcCharsetName); + w2m = ICONV_T_INVALID; } else { - w2m = (iconv_t)-1; + w2m = iconv_open(cname, ms_wcCharsetName.ToAscii()); + if ( w2m == ICONV_T_INVALID ) + { + wxLogTrace(TRACE_STRCONV, + wxT("\"%s\" -> \"%s\" works but not the converse!?"), + ms_wcCharsetName.c_str(), cname.data()); + } } } wxMBConv_iconv::~wxMBConv_iconv() { - if ( m2w != (iconv_t)-1 ) + if ( m2w != ICONV_T_INVALID ) iconv_close(m2w); - if ( w2m != (iconv_t)-1 ) + if ( w2m != ICONV_T_INVALID ) iconv_close(w2m); } 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; @@ -995,7 +1496,8 @@ size_t wxMBConv_iconv::MB2WC(wchar_t *buf, const char *psz, size_t n) const if (ms_wcNeedsSwap) { // convert to native endianness - WC_BSWAP(buf /* _not_ bufPtr */, res) + for ( unsigned i = 0; i < res; i++ ) + buf[n] = WC_BSWAP(buf[i]); } // NB: iconv was given only strlen(psz) characters on input, and so @@ -1025,7 +1527,7 @@ size_t wxMBConv_iconv::MB2WC(wchar_t *buf, const char *psz, size_t n) const if (ICONV_FAILED(cres, inbuf)) { //VS: it is ok if iconv fails, hence trace only - wxLogTrace(wxT("strconv"), wxT("iconv failed: %s"), wxSysErrorMsg(wxSysErrorCode())); + wxLogTrace(TRACE_STRCONV, wxT("iconv failed: %s"), wxSysErrorMsg(wxSysErrorCode())); return (size_t)-1; } @@ -1034,7 +1536,13 @@ size_t wxMBConv_iconv::MB2WC(wchar_t *buf, const char *psz, size_t n) const size_t wxMBConv_iconv::WC2MB(char *buf, const wchar_t *psz, size_t n) const { - size_t inbuf = wxWcslen(psz) * SIZEOF_WCHAR_T; +#if wxUSE_THREADS + // NB: explained in MB2WC + wxMutexLocker lock(wxConstCast(this, wxMBConv_iconv)->m_iconvMutex); +#endif + + size_t inlen = wxWcslen(psz); + size_t inbuf = inlen * SIZEOF_WCHAR_T; size_t outbuf = n; size_t res, cres; @@ -1043,13 +1551,13 @@ size_t wxMBConv_iconv::WC2MB(char *buf, const wchar_t *psz, size_t n) const if (ms_wcNeedsSwap) { // need to copy to temp buffer to switch endianness - // this absolutely doesn't rock! - // (no, doing WC_BSWAP twice on the original buffer won't help, as it + // (doing WC_BSWAP twice on the original buffer won't help, as it // could be in read-only memory, or be accessed in some other thread) - tmpbuf=(wchar_t*)malloc((inbuf+1)*SIZEOF_WCHAR_T); - memcpy(tmpbuf,psz,(inbuf+1)*SIZEOF_WCHAR_T); - WC_BSWAP(tmpbuf, inbuf) - psz=tmpbuf; + 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; } if (buf) @@ -1087,8 +1595,7 @@ size_t wxMBConv_iconv::WC2MB(char *buf, const wchar_t *psz, size_t n) const if (ICONV_FAILED(cres, inbuf)) { - //VS: it is ok if iconv fails, hence trace only - wxLogTrace(wxT("strconv"), wxT("iconv failed: %s"), wxSysErrorMsg(wxSysErrorCode())); + wxLogTrace(TRACE_STRCONV, wxT("iconv failed: %s"), wxSysErrorMsg(wxSysErrorCode())); return (size_t)-1; } @@ -1105,8 +1612,10 @@ size_t wxMBConv_iconv::WC2MB(char *buf, const wchar_t *psz, size_t n) const #ifdef wxHAVE_WIN32_MB2WC // from utils.cpp +#if wxUSE_FONTMAP extern WXDLLIMPEXP_BASE long wxCharsetToCodepage(const wxChar *charset); extern WXDLLIMPEXP_BASE long wxEncodingToCodepage(wxFontEncoding encoding); +#endif class wxMBConv_win32 : public wxMBConv { @@ -1116,6 +1625,7 @@ public: m_CodePage = CP_ACP; } +#if wxUSE_FONTMAP wxMBConv_win32(const wxChar* name) { m_CodePage = wxCharsetToCodepage(name); @@ -1125,52 +1635,805 @@ public: { 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... + // + // Moreover, MB_ERR_INVALID_CHARS is only supported on Win 2K SP4 or + // Win XP or newer and if it is specified on older versions, conversion + // from CP_UTF8 (which can have flags only 0 or MB_ERR_INVALID_CHARS) + // fails. So we can only use the flag on newer Windows versions. + // Additionally, the flag is not supported by UTF7, symbol and CJK + // encodings. See here: + // http://blogs.msdn.com/michkap/archive/2005/04/19/409566.aspx + // http://msdn.microsoft.com/library/en-us/intl/unicode_17si.asp + int flags = 0; + if ( m_CodePage != CP_UTF7 && m_CodePage != CP_SYMBOL && + m_CodePage < 50000 && + IsAtLeastWin2kSP4() ) + { + flags = MB_ERR_INVALID_CHARS; + } + else if ( m_CodePage == CP_UTF8 ) + { + // Avoid round-trip in the special case of UTF-8 by using our + // own UTF-8 conversion code: + return wxMBConvUTF8().MB2WC(buf, psz, n); + } + 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 ); + if ( !len ) + { + // function totally failed + return (size_t)-1; + } + + // if we were really converting and didn't use MB_ERR_INVALID_CHARS, + // check if we succeeded, by doing a double trip: + if ( !flags && buf ) + { + wxCharBuffer mbBuf(n); + if ( ::WideCharToMultiByte + ( + m_CodePage, + 0, + buf, + -1, + mbBuf.data(), + n, + NULL, + NULL + ) == 0 || + strcmp(mbBuf, psz) != 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; + } + } // 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; + return len - 1; } - size_t WC2MB(char *buf, const wchar_t *psz, size_t n) const + size_t WC2MB(char *buf, const wchar_t *pwz, size_t n) const { + /* + we have a problem here: by default, WideCharToMultiByte() may + replace characters unrepresentable in the target code page with bad + quality approximations such as turning "1/2" symbol (U+00BD) into + "1" for the code pages which don't have it and we, obviously, want + to avoid this at any price + + the trouble is that this function does it _silently_, i.e. it won't + even tell us whether it did or not... Win98/2000 and higher provide + WC_NO_BEST_FIT_CHARS but it doesn't work for the older systems and + we have to resort to a round trip, i.e. check that converting back + results in the same string -- this is, of course, expensive but + otherwise we simply can't be sure to not garble the data. + */ + + // determine if we can rely on WC_NO_BEST_FIT_CHARS: according to MSDN + // it doesn't work with CJK encodings (which we test for rather roughly + // here...) nor with UTF-7/8 nor, of course, with Windows versions not + // supporting it + BOOL usedDef wxDUMMY_INITIALIZE(false); + BOOL *pUsedDef; + int flags; + if ( CanUseNoBestFit() && m_CodePage < 50000 ) + { + // it's our lucky day + flags = WC_NO_BEST_FIT_CHARS; + pUsedDef = &usedDef; + } + else // old system or unsupported encoding + { + flags = 0; + pUsedDef = NULL; + } + const size_t len = ::WideCharToMultiByte ( m_CodePage, // code page - 0, // flags (none) - psz, // input string + 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? ); + 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 ? len - 1 : (size_t)-1; + return len - 1; } - bool IsOk() 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; + } + + static bool IsAtLeastWin2kSP4() + { +#ifdef __WXWINCE__ + return false; +#else + static int s_isAtLeastWin2kSP4 = -1; + + if ( s_isAtLeastWin2kSP4 == -1 ) + { + OSVERSIONINFOEX ver; + + memset(&ver, 0, sizeof(ver)); + ver.dwOSVersionInfoSize = sizeof(ver); + GetVersionEx((OSVERSIONINFO*)&ver); + + s_isAtLeastWin2kSP4 = + ((ver.dwMajorVersion > 5) || // Vista+ + (ver.dwMajorVersion == 5 && ver.dwMinorVersion > 0) || // XP/2003 + (ver.dwMajorVersion == 5 && ver.dwMinorVersion == 0 && + ver.wServicePackMajor >= 4)) // 2000 SP4+ + ? 1 : 0; + } + + return s_isAtLeastWin2kSP4 == 1; +#endif + } -public: long m_CodePage; }; #endif // wxHAVE_WIN32_MB2WC +// ============================================================================ +// Cocoa conversion classes +// ============================================================================ + +#if defined(__WXCOCOA__) + +// RN: There is no UTF-32 support in either Core Foundation or +// Cocoa. Strangely enough, internally Core Foundation uses +// UTF 32 internally quite a bit - its just not public (yet). + +#include +#include + +CFStringEncoding wxCFStringEncFromFontEnc(wxFontEncoding encoding) +{ + CFStringEncoding enc = kCFStringEncodingInvalidId ; + if ( encoding == wxFONTENCODING_DEFAULT ) + { + enc = CFStringGetSystemEncoding(); + } + else switch( encoding) + { + case wxFONTENCODING_ISO8859_1 : + enc = kCFStringEncodingISOLatin1 ; + break ; + case wxFONTENCODING_ISO8859_2 : + enc = kCFStringEncodingISOLatin2; + break ; + case wxFONTENCODING_ISO8859_3 : + enc = kCFStringEncodingISOLatin3 ; + break ; + case wxFONTENCODING_ISO8859_4 : + enc = kCFStringEncodingISOLatin4; + break ; + case wxFONTENCODING_ISO8859_5 : + enc = kCFStringEncodingISOLatinCyrillic; + break ; + case wxFONTENCODING_ISO8859_6 : + enc = kCFStringEncodingISOLatinArabic; + break ; + case wxFONTENCODING_ISO8859_7 : + enc = kCFStringEncodingISOLatinGreek; + break ; + case wxFONTENCODING_ISO8859_8 : + enc = kCFStringEncodingISOLatinHebrew; + break ; + case wxFONTENCODING_ISO8859_9 : + enc = kCFStringEncodingISOLatin5; + break ; + case wxFONTENCODING_ISO8859_10 : + enc = kCFStringEncodingISOLatin6; + break ; + case wxFONTENCODING_ISO8859_11 : + enc = kCFStringEncodingISOLatinThai; + break ; + case wxFONTENCODING_ISO8859_13 : + enc = kCFStringEncodingISOLatin7; + break ; + case wxFONTENCODING_ISO8859_14 : + enc = kCFStringEncodingISOLatin8; + break ; + case wxFONTENCODING_ISO8859_15 : + enc = kCFStringEncodingISOLatin9; + break ; + + case wxFONTENCODING_KOI8 : + enc = kCFStringEncodingKOI8_R; + break ; + case wxFONTENCODING_ALTERNATIVE : // MS-DOS CP866 + enc = kCFStringEncodingDOSRussian; + break ; + +// case wxFONTENCODING_BULGARIAN : +// enc = ; +// break ; + + case wxFONTENCODING_CP437 : + enc =kCFStringEncodingDOSLatinUS ; + break ; + case wxFONTENCODING_CP850 : + enc = kCFStringEncodingDOSLatin1; + break ; + case wxFONTENCODING_CP852 : + enc = kCFStringEncodingDOSLatin2; + break ; + case wxFONTENCODING_CP855 : + enc = kCFStringEncodingDOSCyrillic; + break ; + case wxFONTENCODING_CP866 : + enc =kCFStringEncodingDOSRussian ; + break ; + case wxFONTENCODING_CP874 : + enc = kCFStringEncodingDOSThai; + break ; + case wxFONTENCODING_CP932 : + enc = kCFStringEncodingDOSJapanese; + break ; + case wxFONTENCODING_CP936 : + enc =kCFStringEncodingDOSChineseSimplif ; + break ; + case wxFONTENCODING_CP949 : + enc = kCFStringEncodingDOSKorean; + break ; + case wxFONTENCODING_CP950 : + enc = kCFStringEncodingDOSChineseTrad; + break ; + case wxFONTENCODING_CP1250 : + enc = kCFStringEncodingWindowsLatin2; + break ; + case wxFONTENCODING_CP1251 : + enc =kCFStringEncodingWindowsCyrillic ; + break ; + case wxFONTENCODING_CP1252 : + enc =kCFStringEncodingWindowsLatin1 ; + break ; + case wxFONTENCODING_CP1253 : + enc = kCFStringEncodingWindowsGreek; + break ; + case wxFONTENCODING_CP1254 : + enc = kCFStringEncodingWindowsLatin5; + break ; + case wxFONTENCODING_CP1255 : + enc =kCFStringEncodingWindowsHebrew ; + break ; + case wxFONTENCODING_CP1256 : + enc =kCFStringEncodingWindowsArabic ; + break ; + case wxFONTENCODING_CP1257 : + enc = kCFStringEncodingWindowsBalticRim; + break ; +// This only really encodes to UTF7 (if that) evidently +// case wxFONTENCODING_UTF7 : +// enc = kCFStringEncodingNonLossyASCII ; +// break ; + case wxFONTENCODING_UTF8 : + enc = kCFStringEncodingUTF8 ; + break ; + case wxFONTENCODING_EUC_JP : + enc = kCFStringEncodingEUC_JP; + break ; + case wxFONTENCODING_UTF16 : + enc = kCFStringEncodingUnicode ; + break ; + case wxFONTENCODING_MACROMAN : + enc = kCFStringEncodingMacRoman ; + break ; + case wxFONTENCODING_MACJAPANESE : + enc = kCFStringEncodingMacJapanese ; + break ; + case wxFONTENCODING_MACCHINESETRAD : + enc = kCFStringEncodingMacChineseTrad ; + break ; + case wxFONTENCODING_MACKOREAN : + enc = kCFStringEncodingMacKorean ; + break ; + case wxFONTENCODING_MACARABIC : + enc = kCFStringEncodingMacArabic ; + break ; + case wxFONTENCODING_MACHEBREW : + enc = kCFStringEncodingMacHebrew ; + break ; + case wxFONTENCODING_MACGREEK : + enc = kCFStringEncodingMacGreek ; + break ; + case wxFONTENCODING_MACCYRILLIC : + enc = kCFStringEncodingMacCyrillic ; + break ; + case wxFONTENCODING_MACDEVANAGARI : + enc = kCFStringEncodingMacDevanagari ; + break ; + case wxFONTENCODING_MACGURMUKHI : + enc = kCFStringEncodingMacGurmukhi ; + break ; + case wxFONTENCODING_MACGUJARATI : + enc = kCFStringEncodingMacGujarati ; + break ; + case wxFONTENCODING_MACORIYA : + enc = kCFStringEncodingMacOriya ; + break ; + case wxFONTENCODING_MACBENGALI : + enc = kCFStringEncodingMacBengali ; + break ; + case wxFONTENCODING_MACTAMIL : + enc = kCFStringEncodingMacTamil ; + break ; + case wxFONTENCODING_MACTELUGU : + enc = kCFStringEncodingMacTelugu ; + break ; + case wxFONTENCODING_MACKANNADA : + enc = kCFStringEncodingMacKannada ; + break ; + case wxFONTENCODING_MACMALAJALAM : + enc = kCFStringEncodingMacMalayalam ; + break ; + case wxFONTENCODING_MACSINHALESE : + enc = kCFStringEncodingMacSinhalese ; + break ; + case wxFONTENCODING_MACBURMESE : + enc = kCFStringEncodingMacBurmese ; + break ; + case wxFONTENCODING_MACKHMER : + enc = kCFStringEncodingMacKhmer ; + break ; + case wxFONTENCODING_MACTHAI : + enc = kCFStringEncodingMacThai ; + break ; + case wxFONTENCODING_MACLAOTIAN : + enc = kCFStringEncodingMacLaotian ; + break ; + case wxFONTENCODING_MACGEORGIAN : + enc = kCFStringEncodingMacGeorgian ; + break ; + case wxFONTENCODING_MACARMENIAN : + enc = kCFStringEncodingMacArmenian ; + break ; + case wxFONTENCODING_MACCHINESESIMP : + enc = kCFStringEncodingMacChineseSimp ; + break ; + case wxFONTENCODING_MACTIBETAN : + enc = kCFStringEncodingMacTibetan ; + break ; + case wxFONTENCODING_MACMONGOLIAN : + enc = kCFStringEncodingMacMongolian ; + break ; + case wxFONTENCODING_MACETHIOPIC : + enc = kCFStringEncodingMacEthiopic ; + break ; + case wxFONTENCODING_MACCENTRALEUR : + enc = kCFStringEncodingMacCentralEurRoman ; + break ; + case wxFONTENCODING_MACVIATNAMESE : + enc = kCFStringEncodingMacVietnamese ; + break ; + case wxFONTENCODING_MACARABICEXT : + enc = kCFStringEncodingMacExtArabic ; + break ; + case wxFONTENCODING_MACSYMBOL : + enc = kCFStringEncodingMacSymbol ; + break ; + case wxFONTENCODING_MACDINGBATS : + enc = kCFStringEncodingMacDingbats ; + break ; + case wxFONTENCODING_MACTURKISH : + enc = kCFStringEncodingMacTurkish ; + break ; + case wxFONTENCODING_MACCROATIAN : + enc = kCFStringEncodingMacCroatian ; + break ; + case wxFONTENCODING_MACICELANDIC : + enc = kCFStringEncodingMacIcelandic ; + break ; + case wxFONTENCODING_MACROMANIAN : + enc = kCFStringEncodingMacRomanian ; + break ; + case wxFONTENCODING_MACCELTIC : + enc = kCFStringEncodingMacCeltic ; + break ; + case wxFONTENCODING_MACGAELIC : + enc = kCFStringEncodingMacGaelic ; + break ; +// case wxFONTENCODING_MACKEYBOARD : +// enc = kCFStringEncodingMacKeyboardGlyphs ; +// break ; + default : + // because gcc is picky + break ; + } ; + return enc ; +} + +class wxMBConv_cocoa : public wxMBConv +{ +public: + wxMBConv_cocoa() + { + Init(CFStringGetSystemEncoding()) ; + } + +#if wxUSE_FONTMAP + wxMBConv_cocoa(const wxChar* name) + { + Init( wxCFStringEncFromFontEnc(wxFontMapperBase::Get()->CharsetToEncoding(name, false) ) ) ; + } +#endif + + wxMBConv_cocoa(wxFontEncoding encoding) + { + Init( wxCFStringEncFromFontEnc(encoding) ); + } + + ~wxMBConv_cocoa() + { + } + + void Init( CFStringEncoding encoding) + { + m_encoding = encoding ; + } + + size_t MB2WC(wchar_t * szOut, const char * szUnConv, size_t nOutSize) const + { + wxASSERT(szUnConv); + + CFStringRef theString = CFStringCreateWithBytes ( + NULL, //the allocator + (const UInt8*)szUnConv, + strlen(szUnConv), + m_encoding, + false //no BOM/external representation + ); + + wxASSERT(theString); + + size_t nOutLength = CFStringGetLength(theString); + + if (szOut == NULL) + { + CFRelease(theString); + return nOutLength; + } + + CFRange theRange = { 0, nOutSize }; + +#if SIZEOF_WCHAR_T == 4 + UniChar* szUniCharBuffer = new UniChar[nOutSize]; +#endif + + CFStringGetCharacters(theString, theRange, szUniCharBuffer); + + CFRelease(theString); + + szUniCharBuffer[nOutLength] = '\0' ; + +#if SIZEOF_WCHAR_T == 4 + wxMBConvUTF16 converter ; + converter.MB2WC(szOut, (const char*)szUniCharBuffer , nOutSize ) ; + delete[] szUniCharBuffer; +#endif + + return nOutLength; + } + + size_t WC2MB(char *szOut, const wchar_t *szUnConv, size_t nOutSize) const + { + wxASSERT(szUnConv); + + size_t nRealOutSize; + size_t nBufSize = wxWcslen(szUnConv); + UniChar* szUniBuffer = (UniChar*) szUnConv; + +#if SIZEOF_WCHAR_T == 4 + wxMBConvUTF16 converter ; + nBufSize = converter.WC2MB( NULL , szUnConv , 0 ); + szUniBuffer = new UniChar[ (nBufSize / sizeof(UniChar)) + 1] ; + converter.WC2MB( (char*) szUniBuffer , szUnConv, nBufSize + sizeof(UniChar)) ; + nBufSize /= sizeof(UniChar); +#endif + + CFStringRef theString = CFStringCreateWithCharactersNoCopy( + NULL, //allocator + szUniBuffer, + nBufSize, + kCFAllocatorNull //deallocator - we want to deallocate it ourselves + ); + + wxASSERT(theString); + + //Note that CER puts a BOM when converting to unicode + //so we check and use getchars instead in that case + if (m_encoding == kCFStringEncodingUnicode) + { + if (szOut != NULL) + CFStringGetCharacters(theString, CFRangeMake(0, nOutSize - 1), (UniChar*) szOut); + + nRealOutSize = CFStringGetLength(theString) + 1; + } + else + { + CFStringGetBytes( + theString, + CFRangeMake(0, CFStringGetLength(theString)), + m_encoding, + 0, //what to put in characters that can't be converted - + //0 tells CFString to return NULL if it meets such a character + false, //not an external representation + (UInt8*) szOut, + nOutSize, + (CFIndex*) &nRealOutSize + ); + } + + CFRelease(theString); + +#if SIZEOF_WCHAR_T == 4 + delete[] szUniBuffer; +#endif + + return nRealOutSize - 1; + } + + bool IsOk() const + { + return m_encoding != kCFStringEncodingInvalidId && + CFStringIsEncodingAvailable(m_encoding); + } + +private: + CFStringEncoding m_encoding ; +}; + +#endif // defined(__WXCOCOA__) + +// ============================================================================ +// Mac conversion classes +// ============================================================================ + +#if defined(__WXMAC__) && defined(TARGET_CARBON) + +class wxMBConv_mac : public wxMBConv +{ +public: + wxMBConv_mac() + { + Init(CFStringGetSystemEncoding()) ; + } + +#if wxUSE_FONTMAP + wxMBConv_mac(const wxChar* name) + { + Init( wxMacGetSystemEncFromFontEnc(wxFontMapperBase::Get()->CharsetToEncoding(name, false) ) ) ; + } +#endif + + wxMBConv_mac(wxFontEncoding encoding) + { + Init( wxMacGetSystemEncFromFontEnc(encoding) ); + } + + ~wxMBConv_mac() + { + OSStatus status = noErr ; + status = TECDisposeConverter(m_MB2WC_converter); + status = TECDisposeConverter(m_WC2MB_converter); + } + + + 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 + { + OSStatus status = noErr ; + ByteCount byteOutLen ; + ByteCount byteInLen = strlen(psz) ; + 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 + 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 ) ; + 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) + { + //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 + 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; + + //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; + } + } + + 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 @@ -1193,7 +2456,7 @@ public: wxMBConv_wxwin(const wxChar* name) { if (name) - m_enc = wxFontMapper::Get()->CharsetToEncoding(name, false); + m_enc = wxFontMapperBase::Get()->CharsetToEncoding(name, false); else m_enc = wxFONTENCODING_SYSTEM; @@ -1211,7 +2474,10 @@ public: { size_t inbuf = strlen(psz); if (buf) - m2w.Convert(psz,buf); + { + if (!m2w.Convert(psz,buf)) + return (size_t)-1; + } return inbuf; } @@ -1219,7 +2485,10 @@ public: { const size_t inbuf = wxWcslen(psz); if (buf) - w2m.Convert(psz,buf); + { + if (!w2m.Convert(psz,buf)) + return (size_t)-1; + } return inbuf; } @@ -1236,6 +2505,18 @@ public: DECLARE_NO_COPY_CLASS(wxMBConv_wxwin) }; +// make the constructors available for unit testing +WXDLLIMPEXP_BASE wxMBConv* new_wxMBConv_wxwin( const wxChar* name ) +{ + wxMBConv_wxwin* result = new wxMBConv_wxwin( name ); + if ( !result->IsOk() ) + { + delete result; + return 0; + } + return result; +} + #endif // wxUSE_FONTMAP // ============================================================================ @@ -1258,7 +2539,11 @@ wxCSConv::wxCSConv(const wxChar *charset) SetName(charset); } +#if wxUSE_FONTMAP + m_encoding = wxFontMapperBase::GetEncodingFromName(charset); +#else m_encoding = wxFONTENCODING_SYSTEM; +#endif } wxCSConv::wxCSConv(wxFontEncoding encoding) @@ -1317,12 +2602,29 @@ void wxCSConv::SetName(const wxChar *charset) } } +#if wxUSE_FONTMAP +#include "wx/hashmap.h" + +WX_DECLARE_HASH_MAP( wxFontEncoding, wxString, wxIntegerHash, wxIntegerEqual, + wxEncodingNameCache ); + +static wxEncodingNameCache gs_nameCache; +#endif + wxMBConv *wxCSConv::DoCreate() const { +#if wxUSE_FONTMAP + wxLogTrace(TRACE_STRCONV, + wxT("creating conversion for %s"), + (m_name ? m_name + : wxFontMapperBase::GetEncodingName(m_encoding).c_str())); +#endif // wxUSE_FONTMAP + // 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_encoding == wxFONTENCODING_ISO8859_1 || + m_encoding == wxFONTENCODING_DEFAULT ) { // don't convert at all return NULL; @@ -1343,31 +2645,108 @@ wxMBConv *wxCSConv::DoCreate() const #endif // !wxUSE_FONTMAP { wxString name(m_name); + wxFontEncoding encoding(m_encoding); + + if ( !name.empty() ) + { + wxMBConv_iconv *conv = new wxMBConv_iconv(name); + if ( conv->IsOk() ) + return conv; + + delete conv; #if wxUSE_FONTMAP - if ( name.empty() ) - name = wxFontMapper::Get()->GetEncodingName(m_encoding); + encoding = + wxFontMapperBase::Get()->CharsetToEncoding(name, false); #endif // wxUSE_FONTMAP + } +#if wxUSE_FONTMAP + { + const wxEncodingNameCache::iterator it = gs_nameCache.find(encoding); + if ( it != gs_nameCache.end() ) + { + if ( it->second.empty() ) + return NULL; - wxMBConv_iconv *conv = new wxMBConv_iconv(name); - if ( conv->IsOk() ) - return conv; + wxMBConv_iconv *conv = new wxMBConv_iconv(it->second); + if ( conv->IsOk() ) + return conv; - delete conv; + delete conv; + } + + const wxChar** names = wxFontMapperBase::GetAllEncodingNames(encoding); + + for ( ; *names; ++names ) + { + wxMBConv_iconv *conv = new wxMBConv_iconv(*names); + if ( conv->IsOk() ) + { + gs_nameCache[encoding] = *names; + return conv; + } + + delete conv; + } + + gs_nameCache[encoding] = _T(""); // cache the failure + } +#endif // wxUSE_FONTMAP } #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__) + { + // 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); +#else + wxMBConv_mac *conv = new wxMBConv_mac(m_encoding); +#endif + if ( conv->IsOk() ) + return conv; + + delete conv; + } + } +#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; + + delete conv; + } + } +#endif // step (2) wxFontEncoding enc = m_encoding; #if wxUSE_FONTMAP @@ -1376,7 +2755,7 @@ wxMBConv *wxCSConv::DoCreate() const // 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); + enc = wxFontMapperBase::Get()->CharsetToEncoding(m_name, false); } #endif // wxUSE_FONTMAP @@ -1417,15 +2796,28 @@ wxMBConv *wxCSConv::DoCreate() const } #endif // wxUSE_FONTMAP - wxLogError(_("Cannot convert from the charset '%s'!"), - m_name ? m_name + // 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() + wxFontMapperBase::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; } @@ -1441,7 +2833,7 @@ void wxCSConv::CreateConvIfNeeded() const // encoding for this system if ( !m_name && m_encoding == wxFONTENCODING_SYSTEM ) { - self->m_encoding = wxLocale::GetSystemEncoding(); + self->m_name = wxStrdup(wxLocale::GetSystemEncodingName()); } #endif // wxUSE_INTL @@ -1484,7 +2876,7 @@ size_t wxCSConv::WC2MB(char *buf, const wchar_t *psz, size_t n) const { if (psz[c] > 0xFF) return (size_t)-1; - buf[c] = psz[c]; + buf[c] = (char)psz[c]; } } else @@ -1505,6 +2897,8 @@ size_t wxCSConv::WC2MB(char *buf, const wchar_t *psz, size_t n) const #ifdef __WINDOWS__ static wxMBConv_win32 wxConvLibcObj; +#elif defined(__WXMAC__) && !defined(__MACH__) + static wxMBConv_mac wxConvLibcObj ; #else static wxMBConvLibc wxConvLibcObj; #endif @@ -1514,13 +2908,19 @@ static wxCSConv wxConvISO8859_1Obj(wxFONTENCODING_ISO8859_1); static wxMBConvUTF7 wxConvUTF7Obj; static wxMBConvUTF8 wxConvUTF8Obj; - WXDLLIMPEXP_DATA_BASE(wxMBConv&) wxConvLibc = wxConvLibcObj; WXDLLIMPEXP_DATA_BASE(wxCSConv&) wxConvLocal = wxConvLocalObj; WXDLLIMPEXP_DATA_BASE(wxCSConv&) wxConvISO8859_1 = wxConvISO8859_1Obj; WXDLLIMPEXP_DATA_BASE(wxMBConvUTF7&) wxConvUTF7 = wxConvUTF7Obj; WXDLLIMPEXP_DATA_BASE(wxMBConvUTF8&) wxConvUTF8 = wxConvUTF8Obj; WXDLLIMPEXP_DATA_BASE(wxMBConv *) wxConvCurrent = &wxConvLibcObj; +WXDLLIMPEXP_DATA_BASE(wxMBConv *) wxConvFileName = & +#ifdef __WXOSX__ + wxConvUTF8Obj; +#else + wxConvLibcObj; +#endif + #else // !wxUSE_WCHAR_T @@ -1531,5 +2931,3 @@ WXDLLIMPEXP_DATA_BASE(wxMBConv) wxConvLibc, wxConvUTF8; #endif // wxUSE_WCHAR_T/!wxUSE_WCHAR_T - -