// Licence: wxWindows licence
/////////////////////////////////////////////////////////////////////////////
-// ============================================================================
-// declarations
-// ============================================================================
-
-// ----------------------------------------------------------------------------
-// headers
-// ----------------------------------------------------------------------------
-
// For compilers that support precompilation, includes "wx.h".
#include "wx/wxprec.h"
#ifdef __BORLANDC__
- #pragma hdrstop
-#endif
+ #pragma hdrstop
+#endif //__BORLANDC__
#ifndef WX_PRECOMP
#include "wx/intl.h"
#include "wx/log.h"
-#endif // WX_PRECOMP
+ #include "wx/utils.h"
+ #include "wx/hashmap.h"
+#endif
#include "wx/strconv.h"
#if wxUSE_WCHAR_T
-#ifdef __WINDOWS__
- #include "wx/msw/private.h"
- #include "wx/msw/missing.h"
-#endif
-
#ifndef __WXWINCE__
#include <errno.h>
#endif
#include <stdlib.h>
#if defined(__WIN32__) && !defined(__WXMICROWIN__)
+ #include "wx/msw/private.h"
+ #include "wx/msw/missing.h"
#define wxHAVE_WIN32_MB2WC
-#endif // __WIN32__ but !__WXMICROWIN__
+#endif
#ifdef __SALFORDC__
#include <clib.h>
#include "wx/encconv.h"
#include "wx/fontmap.h"
-#include "wx/utils.h"
-#ifdef __WXMAC__
-#ifndef __DARWIN__
-#include <ATSUnicode.h>
-#include <TextCommon.h>
-#include <TextEncodingConverter.h>
-#endif
+#ifdef __DARWIN__
+#include "wx/mac/corefoundation/private/strconv_cf.h"
+#endif //def __DARWIN__
-#include "wx/mac/private.h" // includes mac headers
-#endif
#define TRACE_STRCONV _T("strconv")
+// WC_UTF16 is defined only if sizeof(wchar_t) == 2, otherwise it's supposed to
+// be 4 bytes
#if SIZEOF_WCHAR_T == 2
#define WC_UTF16
#endif
+
// ============================================================================
// implementation
// ============================================================================
+// helper function of cMB2WC(): check if n bytes at this location are all NUL
+static bool NotAllNULs(const char *p, size_t n)
+{
+ while ( n && *p++ == '\0' )
+ n--;
+
+ return n != 0;
+}
+
// ----------------------------------------------------------------------------
-// UTF-16 en/decoding to/from UCS-4
+// UTF-16 en/decoding to/from UCS-4 with surrogates handling
// ----------------------------------------------------------------------------
-
static size_t encode_utf16(wxUint32 input, wxUint16 *output)
{
- if (input<=0xffff)
+ if (input <= 0xffff)
{
if (output)
*output = (wxUint16) input;
+
return 1;
}
- else if (input>=0x110000)
+ else if (input >= 0x110000)
{
- return (size_t)-1;
+ return wxCONV_FAILED;
}
else
{
if (output)
{
- *output++ = (wxUint16) ((input >> 10)+0xd7c0);
- *output = (wxUint16) ((input&0x3ff)+0xdc00);
+ *output++ = (wxUint16) ((input >> 10) + 0xd7c0);
+ *output = (wxUint16) ((input & 0x3ff) + 0xdc00);
}
+
return 2;
}
}
static size_t decode_utf16(const wxUint16* input, wxUint32& output)
{
- if ((*input<0xd800) || (*input>0xdfff))
+ if ((*input < 0xd800) || (*input > 0xdfff))
{
output = *input;
return 1;
}
- else if ((input[1]<0xdc00) || (input[1]>0xdfff))
+ else if ((input[1] < 0xdc00) || (input[1] > 0xdfff))
{
output = *input;
- return (size_t)-1;
+ return wxCONV_FAILED;
}
else
{
}
}
+#ifdef WC_UTF16
+ typedef wchar_t wxDecodeSurrogate_t;
+#else // !WC_UTF16
+ typedef wxUint16 wxDecodeSurrogate_t;
+#endif // WC_UTF16/!WC_UTF16
+
+// returns the next UTF-32 character from the wchar_t buffer and advances the
+// pointer to the character after this one
+//
+// if an invalid character is found, *pSrc is set to NULL, the caller must
+// check for this
+static wxUint32 wxDecodeSurrogate(const wxDecodeSurrogate_t **pSrc)
+{
+ wxUint32 out;
+ const size_t
+ n = decode_utf16(wx_reinterpret_cast(const wxUint16 *, *pSrc), out);
+ if ( n == wxCONV_FAILED )
+ *pSrc = NULL;
+ else
+ *pSrc += n;
+
+ return out;
+}
// ----------------------------------------------------------------------------
// wxMBConv
// ----------------------------------------------------------------------------
-wxMBConv::~wxMBConv()
+size_t
+wxMBConv::ToWChar(wchar_t *dst, size_t dstLen,
+ const char *src, size_t srcLen) const
{
- // nothing to do here (necessary for Darwin linking probably)
-}
+ // although new conversion classes are supposed to implement this function
+ // directly, the existins ones only implement the old MB2WC() and so, to
+ // avoid to have to rewrite all conversion classes at once, we provide a
+ // default (but not efficient) implementation of this one in terms of the
+ // old function by copying the input to ensure that it's NUL-terminated and
+ // then using MB2WC() to convert it
+
+ // the number of chars [which would be] written to dst [if it were not NULL]
+ size_t dstWritten = 0;
+
+ // the number of NULs terminating this string
+ size_t nulLen = 0; // not really needed, but just to avoid warnings
+
+ // if we were not given the input size we just have to assume that the
+ // string is properly terminated as we have no way of knowing how long it
+ // is anyhow, but if we do have the size check whether there are enough
+ // NULs at the end
+ wxCharBuffer bufTmp;
+ const char *srcEnd;
+ if ( srcLen != wxNO_LEN )
+ {
+ // we need to know how to find the end of this string
+ nulLen = GetMBNulLen();
+ if ( nulLen == wxCONV_FAILED )
+ return wxCONV_FAILED;
+
+ // if there are enough NULs we can avoid the copy
+ if ( srcLen < nulLen || NotAllNULs(src + srcLen - nulLen, nulLen) )
+ {
+ // make a copy in order to properly NUL-terminate the string
+ bufTmp = wxCharBuffer(srcLen + nulLen - 1 /* 1 will be added */);
+ char * const p = bufTmp.data();
+ memcpy(p, src, srcLen);
+ for ( char *s = p + srcLen; s < p + srcLen + nulLen; s++ )
+ *s = '\0';
+
+ src = bufTmp;
+ }
-const wxWCharBuffer wxMBConv::cMB2WC(const char *psz) const
-{
- if ( psz )
+ srcEnd = src + srcLen;
+ }
+ else // quit after the first loop iteration
{
- // calculate the length of the buffer needed first
- size_t nLen = MB2WC(NULL, psz, 0);
- if ( nLen != (size_t)-1 )
+ srcEnd = NULL;
+ }
+
+ for ( ;; )
+ {
+ // try to convert the current chunk
+ size_t lenChunk = MB2WC(NULL, src, 0);
+ if ( lenChunk == wxCONV_FAILED )
+ return wxCONV_FAILED;
+
+ lenChunk++; // for the L'\0' at the end of this chunk
+
+ dstWritten += lenChunk;
+
+ if ( lenChunk == 1 )
{
- // now do the actual conversion
- wxWCharBuffer buf(nLen);
- nLen = MB2WC(buf.data(), psz, nLen + 1); // with the trailing NULL
- if ( nLen != (size_t)-1 )
- {
- return buf;
- }
+ // nothing left in the input string, conversion succeeded
+ break;
}
- }
- wxWCharBuffer buf((wchar_t *)NULL);
+ if ( dst )
+ {
+ if ( dstWritten > dstLen )
+ return wxCONV_FAILED;
- return buf;
-}
+ if ( MB2WC(dst, src, lenChunk) == wxCONV_FAILED )
+ return wxCONV_FAILED;
-const wxCharBuffer wxMBConv::cWC2MB(const wchar_t *pwz) const
-{
- if ( pwz )
- {
- size_t nLen = WC2MB(NULL, pwz, 0);
- if ( nLen != (size_t)-1 )
+ dst += lenChunk;
+ }
+
+ if ( !srcEnd )
{
- wxCharBuffer buf(nLen+3); // space for a wxUint32 trailing zero
- nLen = WC2MB(buf.data(), pwz, nLen + 4);
- if ( nLen != (size_t)-1 )
- {
- return buf;
- }
+ // we convert just one chunk in this case as this is the entire
+ // string anyhow
+ break;
}
- }
- wxCharBuffer buf((char *)NULL);
+ // advance the input pointer past the end of this chunk
+ while ( NotAllNULs(src, nulLen) )
+ {
+ // notice that we must skip over multiple bytes here as we suppose
+ // that if NUL takes 2 or 4 bytes, then all the other characters do
+ // too and so if advanced by a single byte we might erroneously
+ // detect sequences of NUL bytes in the middle of the input
+ src += nulLen;
+ }
- return buf;
-}
+ src += nulLen; // skipping over its terminator as well
-const wxWCharBuffer wxMBConv::cMB2WC(const char *szString, size_t nStringLen, size_t* pOutSize) const
-{
- wxASSERT(pOutSize != NULL);
+ // note that ">=" (and not just "==") is needed here as the terminator
+ // we skipped just above could be inside or just after the buffer
+ // delimited by inEnd
+ if ( src >= srcEnd )
+ break;
+ }
- const char* szEnd = szString + nStringLen + 1;
- const char* szPos = szString;
- const char* szStart = szPos;
+ return dstWritten;
+}
- size_t nActualLength = 0;
- size_t nCurrentSize = nStringLen; //try normal size first (should never resize?)
+size_t
+wxMBConv::FromWChar(char *dst, size_t dstLen,
+ const wchar_t *src, size_t srcLen) const
+{
+ // the number of chars [which would be] written to dst [if it were not NULL]
+ size_t dstWritten = 0;
- wxWCharBuffer theBuffer(nCurrentSize);
+ // make a copy of the input string unless it is already properly
+ // NUL-terminated
+ //
+ // if we don't know its length we have no choice but to assume that it is,
+ // indeed, properly terminated
+ wxWCharBuffer bufTmp;
+ if ( srcLen == wxNO_LEN )
+ {
+ srcLen = wxWcslen(src) + 1;
+ }
+ else if ( srcLen != 0 && src[srcLen - 1] != L'\0' )
+ {
+ // make a copy in order to properly NUL-terminate the string
+ bufTmp = wxWCharBuffer(srcLen);
+ memcpy(bufTmp.data(), src, srcLen * sizeof(wchar_t));
+ src = bufTmp;
+ }
- //Convert the string until the length() is reached, continuing the
- //loop every time a null character is reached
- while(szPos != szEnd)
+ const size_t lenNul = GetMBNulLen();
+ for ( const wchar_t * const srcEnd = src + srcLen;
+ src < srcEnd;
+ src += wxWcslen(src) + 1 /* skip L'\0' too */ )
{
- wxASSERT(szPos < szEnd); //something is _really_ screwed up if this rings true
+ // try to convert the current chunk
+ size_t lenChunk = WC2MB(NULL, src, 0);
- //Get the length of the current (sub)string
- size_t nLen = MB2WC(NULL, szPos, 0);
+ if ( lenChunk == wxCONV_FAILED )
+ return wxCONV_FAILED;
- //Invalid conversion?
- if( nLen == (size_t)-1 )
- {
- *pOutSize = 0;
- theBuffer.data()[0u] = wxT('\0');
- return theBuffer;
- }
+ lenChunk += lenNul;
+ dstWritten += lenChunk;
+ if ( dst )
+ {
+ if ( dstWritten > dstLen )
+ return wxCONV_FAILED;
- //Increase the actual length (+1 for current null character)
- nActualLength += nLen + 1;
+ if ( WC2MB(dst, src, lenChunk) == wxCONV_FAILED )
+ return wxCONV_FAILED;
- //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;
+ dst += lenChunk;
}
+ }
- //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;
- }
+ return dstWritten;
+}
- //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;
+size_t wxMBConv::MB2WC(wchar_t *outBuff, const char *inBuff, size_t outLen) const
+{
+ size_t rc = ToWChar(outBuff, outLen, inBuff);
+ if ( rc != wxCONV_FAILED )
+ {
+ // ToWChar() returns the buffer length, i.e. including the trailing
+ // NUL, while this method doesn't take it into account
+ rc--;
}
- //success - return actual length and the buffer
- *pOutSize = nActualLength;
- return theBuffer;
+ return rc;
}
-const wxCharBuffer wxMBConv::cWC2MB(const wchar_t *szString, size_t nStringLen, size_t* pOutSize) const
+size_t wxMBConv::WC2MB(char *outBuff, const wchar_t *inBuff, size_t outLen) const
{
- wxASSERT(pOutSize != NULL);
-
- const wchar_t* szEnd = szString + nStringLen + 1;
- const wchar_t* szPos = szString;
- const wchar_t* szStart = szPos;
+ size_t rc = FromWChar(outBuff, outLen, inBuff);
+ if ( rc != wxCONV_FAILED )
+ {
+ rc -= GetMBNulLen();
+ }
- size_t nActualLength = 0;
- size_t nCurrentSize = nStringLen << 2; //try * 4 first
+ return rc;
+}
- wxCharBuffer theBuffer(nCurrentSize);
+wxMBConv::~wxMBConv()
+{
+ // nothing to do here (necessary for Darwin linking probably)
+}
- //Convert the string until the length() is reached, continuing the
- //loop every time a null character is reached
- while(szPos != szEnd)
+const wxWCharBuffer wxMBConv::cMB2WC(const char *psz) const
+{
+ if ( psz )
{
- wxASSERT(szPos < szEnd); //something is _really_ screwed up if this rings true
+ // calculate the length of the buffer needed first
+ const size_t nLen = ToWChar(NULL, 0, psz);
+ if ( nLen != wxCONV_FAILED )
+ {
+ // now do the actual conversion
+ wxWCharBuffer buf(nLen - 1 /* +1 added implicitly */);
- //Get the length of the current (sub)string
- size_t nLen = WC2MB(NULL, szPos, 0);
+ // +1 for the trailing NULL
+ if ( ToWChar(buf.data(), nLen, psz) != wxCONV_FAILED )
+ return buf;
+ }
+ }
- //Invalid conversion?
- if( nLen == (size_t)-1 )
+ return wxWCharBuffer();
+}
+
+const wxCharBuffer wxMBConv::cWC2MB(const wchar_t *pwz) const
+{
+ if ( pwz )
+ {
+ const size_t nLen = FromWChar(NULL, 0, pwz);
+ if ( nLen != wxCONV_FAILED )
{
- *pOutSize = 0;
- theBuffer.data()[0u] = wxT('\0');
- return theBuffer;
+ wxCharBuffer buf(nLen - 1);
+ if ( FromWChar(buf.data(), nLen, pwz) != wxCONV_FAILED )
+ return buf;
}
+ }
- //Increase the actual length (+1 for current null character)
- nActualLength += nLen + 1;
+ return wxCharBuffer();
+}
- //if buffer too big, realloc the buffer
- if (nActualLength > (nCurrentSize+1))
+const wxWCharBuffer
+wxMBConv::cMB2WC(const char *inBuff, size_t inLen, size_t *outLen) const
+{
+ const size_t dstLen = ToWChar(NULL, 0, inBuff, inLen);
+ if ( dstLen != wxCONV_FAILED )
+ {
+ wxWCharBuffer wbuf(dstLen - 1);
+ if ( ToWChar(wbuf.data(), dstLen, inBuff, inLen) != wxCONV_FAILED )
{
- wxCharBuffer theNewBuffer(nCurrentSize << 1);
- memcpy(theNewBuffer.data(), theBuffer.data(), nCurrentSize);
- theBuffer = theNewBuffer;
- nCurrentSize <<= 1;
+ if ( outLen )
+ {
+ *outLen = dstLen;
+ if ( wbuf[dstLen - 1] == L'\0' )
+ (*outLen)--;
+ }
+
+ return wbuf;
}
+ }
- //Convert the current (sub)string
- if(WC2MB(&theBuffer.data()[szPos - szStart], szPos, nLen + 1) == (size_t)-1 )
+ if ( outLen )
+ *outLen = 0;
+
+ return wxWCharBuffer();
+}
+
+const wxCharBuffer
+wxMBConv::cWC2MB(const wchar_t *inBuff, size_t inLen, size_t *outLen) const
+{
+ size_t dstLen = FromWChar(NULL, 0, inBuff, inLen);
+ if ( dstLen != wxCONV_FAILED )
+ {
+ // special case of empty input: can't allocate 0 size buffer below as
+ // wxCharBuffer insists on NUL-terminating it
+ wxCharBuffer buf(dstLen ? dstLen - 1 : 1);
+ if ( FromWChar(buf.data(), dstLen, inBuff, inLen) != wxCONV_FAILED )
{
- *pOutSize = 0;
- theBuffer.data()[0u] = wxT('\0');
- return theBuffer;
- }
+ if ( outLen )
+ {
+ *outLen = dstLen;
+
+ const size_t nulLen = GetMBNulLen();
+ if ( dstLen >= nulLen &&
+ !NotAllNULs(buf.data() + dstLen - nulLen, nulLen) )
+ {
+ // in this case the output is NUL-terminated and we're not
+ // supposed to count NUL
+ *outLen -= nulLen;
+ }
+ }
- //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;
+ return buf;
+ }
}
- //success - return actual length and the buffer
- *pOutSize = nActualLength;
- return theBuffer;
+ if ( outLen )
+ *outLen = 0;
+
+ return wxCharBuffer();
}
// ----------------------------------------------------------------------------
return wxWC2MB(buf, psz, n);
}
-#ifdef __UNIX__
-
// ----------------------------------------------------------------------------
// wxConvBrokenFileNames
// ----------------------------------------------------------------------------
-wxConvBrokenFileNames::wxConvBrokenFileNames(const wxChar *charset)
+#ifdef __UNIX__
+
+wxConvBrokenFileNames::wxConvBrokenFileNames(const wxString& charset)
{
- if ( !charset || wxStricmp(charset, _T("UTF-8")) == 0
- || wxStricmp(charset, _T("UTF8")) == 0 )
- m_conv = new wxMBConvUTF8(wxMBConvUTF8::MAP_INVALID_UTF8_TO_OCTAL);
+ if ( wxStricmp(charset, _T("UTF-8")) == 0 ||
+ wxStricmp(charset, _T("UTF8")) == 0 )
+ m_conv = new wxMBConvUTF8(wxMBConvUTF8::MAP_INVALID_UTF8_TO_PUA);
else
m_conv = new wxCSConv(charset);
}
-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
+#endif // __UNIX__
// ----------------------------------------------------------------------------
// UTF-7
if ( !ok )
{
// in valid UTF7 we should have valid characters after '+'
- return (size_t)-1;
+ return wxCONV_FAILED;
}
if (*psz == '-')
size_t wxMBConvUTF7::WC2MB(char *buf, const wchar_t *psz, size_t n) const
{
-
-
size_t len = 0;
while (*psz && ((!buf) || (len < n)))
// plain ASCII char
if (buf)
*buf++ = (char)cc;
+
len++;
}
#ifndef WC_UTF16
else if (((wxUint32)cc) > 0xffff)
{
// no surrogate pair generation (yet?)
- return (size_t)-1;
+ return wxCONV_FAILED;
}
#endif
else
{
if (buf)
*buf++ = '+';
+
len++;
if (cc != '+')
{
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;
}
const wxUint32 wxUnicodePUA = 0x100000;
const wxUint32 wxUnicodePUAEnd = wxUnicodePUA + 256;
+// this table gives the length of the UTF-8 encoding from its first character:
+unsigned char tableUtf8Lengths[256] = {
+ // single-byte sequences (ASCII):
+ 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, // 00..0F
+ 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, // 10..1F
+ 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, // 20..2F
+ 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, // 30..3F
+ 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, // 40..4F
+ 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, // 50..5F
+ 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, // 60..6F
+ 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, // 70..7F
+
+ // these are invalid:
+ 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, // 80..8F
+ 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, // 90..9F
+ 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, // A0..AF
+ 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, // B0..BF
+ 0, 0, // C0,C1
+
+ // two-byte sequences:
+ 2, 2, 2, 2, 2, 2, 2, 2, 2, 2, 2, 2, 2, 2, // C2..CF
+ 2, 2, 2, 2, 2, 2, 2, 2, 2, 2, 2, 2, 2, 2, 2, 2, // D0..DF
+
+ // three-byte sequences:
+ 3, 3, 3, 3, 3, 3, 3, 3, 3, 3, 3, 3, 3, 3, 3, 3, // E0..EF
+
+ // four-byte sequences:
+ 4, 4, 4, 4, 4, // F0..F4
+
+ // these are invalid again (5- or 6-byte
+ // sequences and sequences for code points
+ // above U+10FFFF, as restricted by RFC 3629):
+ 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0 // F5..FF
+};
+
+size_t
+wxMBConvStrictUTF8::ToWChar(wchar_t *dst, size_t dstLen,
+ const char *src, size_t srcLen) const
+{
+ wchar_t *out = dstLen ? dst : NULL;
+ size_t written = 0;
+
+ if ( srcLen == wxNO_LEN )
+ srcLen = strlen(src) + 1;
+
+ for ( const char *p = src; ; p++ )
+ {
+ if ( !(srcLen == wxNO_LEN ? *p : srcLen) )
+ {
+ // all done successfully, just add the trailing NULL if we are not
+ // using explicit length
+ if ( srcLen == wxNO_LEN )
+ {
+ if ( out )
+ {
+ if ( !dstLen )
+ break;
+
+ *out = L'\0';
+ }
+
+ written++;
+ }
+
+ return written;
+ }
+
+ unsigned char c = *p;
+ unsigned len = tableUtf8Lengths[c];
+ if ( !len )
+ break;
+
+ if ( srcLen < len ) // the test works for wxNO_LEN too
+ break;
+
+ if ( srcLen != wxNO_LEN )
+ srcLen -= len;
+
+ if ( out && !dstLen-- )
+ break;
+
+
+ // Char. number range | UTF-8 octet sequence
+ // (hexadecimal) | (binary)
+ // ----------------------+---------------------------------------------
+ // 0000 0000 - 0000 007F | 0xxxxxxx
+ // 0000 0080 - 0000 07FF | 110xxxxx 10xxxxxx
+ // 0000 0800 - 0000 FFFF | 1110xxxx 10xxxxxx 10xxxxxx
+ // 0001 0000 - 0010 FFFF | 11110xxx 10xxxxxx 10xxxxxx 10xxxxxx
+ //
+ // Code point value is stored in bits marked with 'x', lowest-order bit
+ // of the value on the right side in the diagram above.
+ // (from RFC 3629)
+
+ // mask to extract lead byte's value ('x' bits above), by sequence length:
+ static const unsigned char leadValueMask[] = { 0x7F, 0x1F, 0x0F, 0x07 };
+
+ // mask and value of lead byte's most significant bits, by length:
+ static const unsigned char leadMarkerMask[] = { 0x80, 0xE0, 0xF0, 0xF8 };
+ static const unsigned char leadMarkerVal[] = { 0x00, 0xC0, 0xE0, 0xF0 };
+
+ len--; // it's more convenient to work with 0-based length here
+
+ // extract the lead byte's value bits:
+ if ( (c & leadMarkerMask[len]) != leadMarkerVal[len] )
+ break;
+
+ wxUint32 code = c & leadValueMask[len];
+
+ // all remaining bytes, if any, are handled in the same way regardless of
+ // sequence's length:
+ for ( ; len; --len )
+ {
+ c = *++p;
+ if ( (c & 0xC0) != 0x80 )
+ return wxCONV_FAILED;
+
+ code <<= 6;
+ code |= c & 0x3F;
+ }
+
+#ifdef WC_UTF16
+ // cast is ok because wchar_t == wxUint16 if WC_UTF16
+ if ( encode_utf16(code, (wxUint16 *)out) == 2 )
+ {
+ if ( out )
+ out++;
+ written++;
+ }
+#else // !WC_UTF16
+ if ( out )
+ *out = code;
+#endif // WC_UTF16/!WC_UTF16
+
+ if ( out )
+ out++;
+
+ written++;
+ }
+
+ return wxCONV_FAILED;
+}
+
+size_t
+wxMBConvStrictUTF8::FromWChar(char *dst, size_t dstLen,
+ const wchar_t *src, size_t srcLen) const
+{
+ char *out = dstLen ? dst : NULL;
+ size_t written = 0;
+
+ for ( const wchar_t *wp = src; ; wp++ )
+ {
+ if ( !(srcLen == wxNO_LEN ? *wp : srcLen--) )
+ {
+ // all done successfully, just add the trailing NULL if we are not
+ // using explicit length
+ if ( srcLen == wxNO_LEN )
+ {
+ if ( out )
+ {
+ if ( !dstLen )
+ break;
+
+ *out = '\0';
+ }
+
+ written++;
+ }
+
+ return written;
+ }
+
+
+ wxUint32 code;
+#ifdef WC_UTF16
+ // cast is ok for WC_UTF16
+ if ( decode_utf16((const wxUint16 *)wp, code) == 2 )
+ {
+ // skip the next char too as we decoded a surrogate
+ wp++;
+ }
+#else // wchar_t is UTF-32
+ code = *wp & 0x7fffffff;
+#endif
+
+ unsigned len;
+ if ( code <= 0x7F )
+ {
+ len = 1;
+ if ( out )
+ {
+ if ( dstLen < len )
+ break;
+
+ out[0] = (char)code;
+ }
+ }
+ else if ( code <= 0x07FF )
+ {
+ len = 2;
+ if ( out )
+ {
+ if ( dstLen < len )
+ break;
+
+ // NB: this line takes 6 least significant bits, encodes them as
+ // 10xxxxxx and discards them so that the next byte can be encoded:
+ out[1] = 0x80 | (code & 0x3F); code >>= 6;
+ out[0] = 0xC0 | code;
+ }
+ }
+ else if ( code < 0xFFFF )
+ {
+ len = 3;
+ if ( out )
+ {
+ if ( dstLen < len )
+ break;
+
+ out[2] = 0x80 | (code & 0x3F); code >>= 6;
+ out[1] = 0x80 | (code & 0x3F); code >>= 6;
+ out[0] = 0xE0 | code;
+ }
+ }
+ else if ( code <= 0x10FFFF )
+ {
+ len = 4;
+ if ( out )
+ {
+ if ( dstLen < len )
+ break;
+
+ out[3] = 0x80 | (code & 0x3F); code >>= 6;
+ out[2] = 0x80 | (code & 0x3F); code >>= 6;
+ out[1] = 0x80 | (code & 0x3F); code >>= 6;
+ out[0] = 0xF0 | code;
+ }
+ }
+ else
+ {
+ wxFAIL_MSG( _T("trying to encode undefined Unicode character") );
+ break;
+ }
+
+ if ( out )
+ {
+ out += len;
+ dstLen -= len;
+ }
+
+ written += len;
+ }
+
+ // we only get here if an error occurs during decoding
+ return wxCONV_FAILED;
+}
+
size_t wxMBConvUTF8::MB2WC(wchar_t *buf, const char *psz, size_t n) const
{
+ if ( m_options == MAP_INVALID_UTF8_NOT )
+ return wxMBConvStrictUTF8::MB2WC(buf, psz, n);
+
size_t len = 0;
while (*psz && ((!buf) || (len < n)))
unsigned cnt;
for (cnt = 0; fc & 0x80; cnt++)
fc <<= 1;
+
if (!cnt)
{
// plain ASCII char
invalid = true;
break;
}
+
psz++;
res = (res << 6) | (cc & 0x3f);
}
+
if (invalid || res <= utf8_max[ocnt])
{
// illegal UTF-8 encoding
else
{
#ifdef WC_UTF16
- // cast is ok because wchar_t == wxUuint16 if WC_UTF16
+ // cast is ok because wchar_t == wxUint16 if WC_UTF16
size_t pa = encode_utf16(res, (wxUint16 *)buf);
- if (pa == (size_t)-1)
+ if (pa == wxCONV_FAILED)
{
invalid = true;
}
#endif // WC_UTF16/!WC_UTF16
}
}
+
if (invalid)
{
if (m_options & MAP_INVALID_UTF8_TO_PUA)
#ifdef WC_UTF16
// cast is ok because wchar_t == wxUuint16 if WC_UTF16
size_t pa = encode_utf16((unsigned char)*opsz + wxUnicodePUA, (wxUint16 *)buf);
- wxASSERT(pa != (size_t)-1);
+ wxASSERT(pa != wxCONV_FAILED);
if (buf)
buf += pa;
opsz++;
*buf++ = (wchar_t)( L'0' + (on % 0100) / 010 );
*buf++ = (wchar_t)( L'0' + on % 010 );
}
+
opsz++;
len += 4;
}
}
else // MAP_INVALID_UTF8_NOT
{
- return (size_t)-1;
+ return wxCONV_FAILED;
}
}
}
}
+
if (buf && (len < n))
*buf = 0;
+
return len;
}
size_t wxMBConvUTF8::WC2MB(char *buf, const wchar_t *psz, size_t n) const
{
+ if ( m_options == MAP_INVALID_UTF8_NOT )
+ return wxMBConvStrictUTF8::WC2MB(buf, psz, n);
+
size_t len = 0;
while (*psz && ((!buf) || (len < n)))
{
wxUint32 cc;
+
#ifdef WC_UTF16
// cast is ok for WC_UTF16
size_t pa = decode_utf16((const wxUint16 *)psz, cc);
- psz += (pa == (size_t)-1) ? 1 : pa;
+ psz += (pa == wxCONV_FAILED) ? 1 : pa;
#else
- cc=(*psz++) & 0x7fffffff;
+ cc = (*psz++) & 0x7fffffff;
#endif
if ( (m_options & MAP_INVALID_UTF8_TO_PUA)
{
if (buf)
{
- *buf++ = (char) ((psz[0] - L'0')*0100 +
- (psz[1] - L'0')*010 +
+ *buf++ = (char) ((psz[0] - L'0') * 0100 +
+ (psz[1] - L'0') * 010 +
(psz[2] - L'0'));
}
else
{
unsigned cnt;
- for (cnt = 0; cc > utf8_max[cnt]; cnt++) {}
+ for (cnt = 0; cc > utf8_max[cnt]; cnt++)
+ {
+ }
+
if (!cnt)
{
// plain ASCII char
*buf++ = (char) cc;
len++;
}
-
else
{
len += cnt + 1;
}
}
- if (buf && (len<n))
+ if (buf && (len < n))
*buf = 0;
return len;
}
-// ----------------------------------------------------------------------------
+// ============================================================================
// UTF-16
-// ----------------------------------------------------------------------------
+// ============================================================================
#ifdef WORDS_BIGENDIAN
#define wxMBConvUTF16straight wxMBConvUTF16BE
#define wxMBConvUTF16straight wxMBConvUTF16LE
#endif
+/* static */
+size_t wxMBConvUTF16Base::GetLength(const char *src, size_t srcLen)
+{
+ if ( srcLen == wxNO_LEN )
+ {
+ // count the number of bytes in input, including the trailing NULs
+ const wxUint16 *inBuff = wx_reinterpret_cast(const wxUint16 *, src);
+ for ( srcLen = 1; *inBuff++; srcLen++ )
+ ;
+
+ srcLen *= BYTES_PER_CHAR;
+ }
+ else // we already have the length
+ {
+ // we can only convert an entire number of UTF-16 characters
+ if ( srcLen % BYTES_PER_CHAR )
+ return wxCONV_FAILED;
+ }
+
+ return srcLen;
+}
+// case when in-memory representation is UTF-16 too
#ifdef WC_UTF16
-// copy 16bit MB to 16bit String
-size_t wxMBConvUTF16straight::MB2WC(wchar_t *buf, const char *psz, size_t n) const
+// ----------------------------------------------------------------------------
+// conversions without endianness change
+// ----------------------------------------------------------------------------
+
+size_t
+wxMBConvUTF16straight::ToWChar(wchar_t *dst, size_t dstLen,
+ const char *src, size_t srcLen) const
{
- size_t len=0;
+ // set up the scene for using memcpy() (which is presumably more efficient
+ // than copying the bytes one by one)
+ srcLen = GetLength(src, srcLen);
+ if ( srcLen == wxNO_LEN )
+ return wxCONV_FAILED;
- while (*(wxUint16*)psz && (!buf || len < n))
+ const size_t inLen = srcLen / BYTES_PER_CHAR;
+ if ( dst )
{
- if (buf)
- *buf++ = *(wxUint16*)psz;
- len++;
+ if ( dstLen < inLen )
+ return wxCONV_FAILED;
- psz += sizeof(wxUint16);
+ memcpy(dst, src, srcLen);
}
- if (buf && len<n) *buf=0;
- return len;
+ return inLen;
}
-
-// copy 16bit String to 16bit MB
-size_t wxMBConvUTF16straight::WC2MB(char *buf, const wchar_t *psz, size_t n) const
+size_t
+wxMBConvUTF16straight::FromWChar(char *dst, size_t dstLen,
+ const wchar_t *src, size_t srcLen) const
{
- size_t len=0;
+ if ( srcLen == wxNO_LEN )
+ srcLen = wxWcslen(src) + 1;
- while (*psz && (!buf || len < n))
+ srcLen *= BYTES_PER_CHAR;
+
+ if ( dst )
{
- if (buf)
- {
- *(wxUint16*)buf = *psz;
- buf += sizeof(wxUint16);
- }
- len += sizeof(wxUint16);
- psz++;
+ if ( dstLen < srcLen )
+ return wxCONV_FAILED;
+
+ memcpy(dst, src, srcLen);
}
- if (buf && len<=n-sizeof(wxUint16)) *(wxUint16*)buf=0;
- return len;
+ return srcLen;
}
+// ----------------------------------------------------------------------------
+// endian-reversing conversions
+// ----------------------------------------------------------------------------
-// swap 16bit MB to 16bit String
-size_t wxMBConvUTF16swap::MB2WC(wchar_t *buf, const char *psz, size_t n) const
+size_t
+wxMBConvUTF16swap::ToWChar(wchar_t *dst, size_t dstLen,
+ const char *src, size_t srcLen) const
{
- size_t len=0;
+ srcLen = GetLength(src, srcLen);
+ if ( srcLen == wxNO_LEN )
+ return wxCONV_FAILED;
+
+ srcLen /= BYTES_PER_CHAR;
- while (*(wxUint16*)psz && (!buf || len < n))
+ if ( dst )
{
- if (buf)
+ if ( dstLen < srcLen )
+ return wxCONV_FAILED;
+
+ const wxUint16 *inBuff = wx_reinterpret_cast(const wxUint16 *, src);
+ for ( size_t n = 0; n < srcLen; n++, inBuff++ )
{
- ((char *)buf)[0] = psz[1];
- ((char *)buf)[1] = psz[0];
- buf++;
+ *dst++ = wxUINT16_SWAP_ALWAYS(*inBuff);
}
- len++;
- psz += sizeof(wxUint16);
}
- if (buf && len<n) *buf=0;
- return len;
+ return srcLen;
}
-
-// swap 16bit MB to 16bit String
-size_t wxMBConvUTF16swap::WC2MB(char *buf, const wchar_t *psz, size_t n) const
+size_t
+wxMBConvUTF16swap::FromWChar(char *dst, size_t dstLen,
+ const wchar_t *src, size_t srcLen) const
{
- size_t len=0;
+ if ( srcLen == wxNO_LEN )
+ srcLen = wxWcslen(src) + 1;
+
+ srcLen *= BYTES_PER_CHAR;
- while (*psz && (!buf || len < n))
+ if ( dst )
{
- if (buf)
+ if ( dstLen < srcLen )
+ return wxCONV_FAILED;
+
+ wxUint16 *outBuff = wx_reinterpret_cast(wxUint16 *, dst);
+ for ( size_t n = 0; n < srcLen; n += BYTES_PER_CHAR, src++ )
{
- *buf++ = ((char*)psz)[1];
- *buf++ = ((char*)psz)[0];
+ *outBuff++ = wxUINT16_SWAP_ALWAYS(*src);
}
- len += sizeof(wxUint16);
- psz++;
}
- if (buf && len<=n-sizeof(wxUint16)) *(wxUint16*)buf=0;
- return len;
+ return srcLen;
}
+#else // !WC_UTF16: wchar_t is UTF-32
-#else // WC_UTF16
-
+// ----------------------------------------------------------------------------
+// conversions without endianness change
+// ----------------------------------------------------------------------------
-// copy 16bit MB to 32bit String
-size_t wxMBConvUTF16straight::MB2WC(wchar_t *buf, const char *psz, size_t n) const
+size_t
+wxMBConvUTF16straight::ToWChar(wchar_t *dst, size_t dstLen,
+ const char *src, size_t srcLen) const
{
- size_t len=0;
+ srcLen = GetLength(src, srcLen);
+ if ( srcLen == wxNO_LEN )
+ return wxCONV_FAILED;
- while (*(wxUint16*)psz && (!buf || len < n))
+ const size_t inLen = srcLen / BYTES_PER_CHAR;
+ if ( !dst )
{
- wxUint32 cc;
- size_t pa=decode_utf16((wxUint16*)psz, cc);
- if (pa == (size_t)-1)
- return pa;
+ // optimization: return maximal space which could be needed for this
+ // string even if the real size could be smaller if the buffer contains
+ // any surrogates
+ return inLen;
+ }
- if (buf)
- *buf++ = (wchar_t)cc;
- len++;
- psz += pa * sizeof(wxUint16);
+ size_t outLen = 0;
+ const wxUint16 *inBuff = wx_reinterpret_cast(const wxUint16 *, src);
+ for ( const wxUint16 * const inEnd = inBuff + inLen; inBuff < inEnd; )
+ {
+ const wxUint32 ch = wxDecodeSurrogate(&inBuff);
+ if ( !inBuff )
+ return wxCONV_FAILED;
+
+ if ( ++outLen > dstLen )
+ return wxCONV_FAILED;
+
+ *dst++ = ch;
}
- if (buf && len<n) *buf=0;
- return len;
-}
+ return outLen;
+}
-// copy 32bit String to 16bit MB
-size_t wxMBConvUTF16straight::WC2MB(char *buf, const wchar_t *psz, size_t n) const
+size_t
+wxMBConvUTF16straight::FromWChar(char *dst, size_t dstLen,
+ const wchar_t *src, size_t srcLen) const
{
- size_t len=0;
+ if ( srcLen == wxNO_LEN )
+ srcLen = wxWcslen(src) + 1;
- while (*psz && (!buf || len < n))
+ size_t outLen = 0;
+ wxUint16 *outBuff = wx_reinterpret_cast(wxUint16 *, dst);
+ for ( size_t n = 0; n < srcLen; n++ )
{
wxUint16 cc[2];
- size_t pa=encode_utf16(*psz, cc);
-
- if (pa == (size_t)-1)
- return pa;
+ const size_t numChars = encode_utf16(*src++, cc);
+ if ( numChars == wxCONV_FAILED )
+ return wxCONV_FAILED;
- if (buf)
+ outLen += numChars * BYTES_PER_CHAR;
+ if ( outBuff )
{
- *(wxUint16*)buf = cc[0];
- buf += sizeof(wxUint16);
- if (pa > 1)
+ if ( outLen > dstLen )
+ return wxCONV_FAILED;
+
+ *outBuff++ = cc[0];
+ if ( numChars == 2 )
{
- *(wxUint16*)buf = cc[1];
- buf += sizeof(wxUint16);
+ // second character of a surrogate
+ *outBuff++ = cc[1];
}
}
-
- len += pa*sizeof(wxUint16);
- psz++;
}
- if (buf && len<=n-sizeof(wxUint16)) *(wxUint16*)buf=0;
- return len;
+ return outLen;
}
+// ----------------------------------------------------------------------------
+// endian-reversing conversions
+// ----------------------------------------------------------------------------
-// swap 16bit MB to 32bit String
-size_t wxMBConvUTF16swap::MB2WC(wchar_t *buf, const char *psz, size_t n) const
+size_t
+wxMBConvUTF16swap::ToWChar(wchar_t *dst, size_t dstLen,
+ const char *src, size_t srcLen) const
{
- size_t len=0;
+ srcLen = GetLength(src, srcLen);
+ if ( srcLen == wxNO_LEN )
+ return wxCONV_FAILED;
- while (*(wxUint16*)psz && (!buf || len < n))
+ const size_t inLen = srcLen / BYTES_PER_CHAR;
+ if ( !dst )
{
- wxUint32 cc;
- char tmp[4];
- tmp[0]=psz[1]; tmp[1]=psz[0];
- tmp[2]=psz[3]; tmp[3]=psz[2];
+ // optimization: return maximal space which could be needed for this
+ // string even if the real size could be smaller if the buffer contains
+ // any surrogates
+ return inLen;
+ }
+
+ size_t outLen = 0;
+ const wxUint16 *inBuff = wx_reinterpret_cast(const wxUint16 *, src);
+ for ( const wxUint16 * const inEnd = inBuff + inLen; inBuff < inEnd; )
+ {
+ wxUint32 ch;
+ wxUint16 tmp[2];
- size_t pa=decode_utf16((wxUint16*)tmp, cc);
- if (pa == (size_t)-1)
- return pa;
+ tmp[0] = wxUINT16_SWAP_ALWAYS(*inBuff);
+ inBuff++;
+ tmp[1] = wxUINT16_SWAP_ALWAYS(*inBuff);
- if (buf)
- *buf++ = (wchar_t)cc;
+ const size_t numChars = decode_utf16(tmp, ch);
+ if ( numChars == wxCONV_FAILED )
+ return wxCONV_FAILED;
+
+ if ( numChars == 2 )
+ inBuff++;
+
+ if ( ++outLen > dstLen )
+ return wxCONV_FAILED;
- len++;
- psz += pa * sizeof(wxUint16);
+ *dst++ = ch;
}
- if (buf && len<n) *buf=0;
- return len;
-}
+ return outLen;
+}
-// swap 32bit String to 16bit MB
-size_t wxMBConvUTF16swap::WC2MB(char *buf, const wchar_t *psz, size_t n) const
+size_t
+wxMBConvUTF16swap::FromWChar(char *dst, size_t dstLen,
+ const wchar_t *src, size_t srcLen) const
{
- size_t len=0;
+ if ( srcLen == wxNO_LEN )
+ srcLen = wxWcslen(src) + 1;
- while (*psz && (!buf || len < n))
+ size_t outLen = 0;
+ wxUint16 *outBuff = wx_reinterpret_cast(wxUint16 *, dst);
+ for ( const wchar_t *srcEnd = src + srcLen; src < srcEnd; src++ )
{
wxUint16 cc[2];
- size_t pa=encode_utf16(*psz, cc);
+ const size_t numChars = encode_utf16(*src, cc);
+ if ( numChars == wxCONV_FAILED )
+ return wxCONV_FAILED;
- if (pa == (size_t)-1)
- return pa;
-
- if (buf)
+ outLen += numChars * BYTES_PER_CHAR;
+ if ( outBuff )
{
- *buf++ = ((char*)cc)[1];
- *buf++ = ((char*)cc)[0];
- if (pa > 1)
+ if ( outLen > dstLen )
+ return wxCONV_FAILED;
+
+ *outBuff++ = wxUINT16_SWAP_ALWAYS(cc[0]);
+ if ( numChars == 2 )
{
- *buf++ = ((char*)cc)[3];
- *buf++ = ((char*)cc)[2];
+ // second character of a surrogate
+ *outBuff++ = wxUINT16_SWAP_ALWAYS(cc[1]);
}
}
-
- len += pa*sizeof(wxUint16);
- psz++;
}
- if (buf && len<=n-sizeof(wxUint16)) *(wxUint16*)buf=0;
- return len;
+ return outLen;
}
-#endif // WC_UTF16
+#endif // WC_UTF16/!WC_UTF16
-// ----------------------------------------------------------------------------
+// ============================================================================
// UTF-32
-// ----------------------------------------------------------------------------
+// ============================================================================
#ifdef WORDS_BIGENDIAN
-#define wxMBConvUTF32straight wxMBConvUTF32BE
-#define wxMBConvUTF32swap wxMBConvUTF32LE
+ #define wxMBConvUTF32straight wxMBConvUTF32BE
+ #define wxMBConvUTF32swap wxMBConvUTF32LE
#else
-#define wxMBConvUTF32swap wxMBConvUTF32BE
-#define wxMBConvUTF32straight wxMBConvUTF32LE
+ #define wxMBConvUTF32swap wxMBConvUTF32BE
+ #define wxMBConvUTF32straight wxMBConvUTF32LE
#endif
WXDLLIMPEXP_DATA_BASE(wxMBConvUTF32LE) wxConvUTF32LE;
WXDLLIMPEXP_DATA_BASE(wxMBConvUTF32BE) wxConvUTF32BE;
+/* static */
+size_t wxMBConvUTF32Base::GetLength(const char *src, size_t srcLen)
+{
+ if ( srcLen == wxNO_LEN )
+ {
+ // count the number of bytes in input, including the trailing NULs
+ const wxUint32 *inBuff = wx_reinterpret_cast(const wxUint32 *, src);
+ for ( srcLen = 1; *inBuff++; srcLen++ )
+ ;
+
+ srcLen *= BYTES_PER_CHAR;
+ }
+ else // we already have the length
+ {
+ // we can only convert an entire number of UTF-32 characters
+ if ( srcLen % BYTES_PER_CHAR )
+ return wxCONV_FAILED;
+ }
+
+ return srcLen;
+}
+// case when in-memory representation is UTF-16
#ifdef WC_UTF16
-// copy 32bit MB to 16bit String
-size_t wxMBConvUTF32straight::MB2WC(wchar_t *buf, const char *psz, size_t n) const
+// ----------------------------------------------------------------------------
+// conversions without endianness change
+// ----------------------------------------------------------------------------
+
+size_t
+wxMBConvUTF32straight::ToWChar(wchar_t *dst, size_t dstLen,
+ const char *src, size_t srcLen) const
{
- size_t len=0;
+ srcLen = GetLength(src, srcLen);
+ if ( srcLen == wxNO_LEN )
+ return wxCONV_FAILED;
- while (*(wxUint32*)psz && (!buf || len < n))
+ const wxUint32 *inBuff = wx_reinterpret_cast(const wxUint32 *, src);
+ const size_t inLen = srcLen / BYTES_PER_CHAR;
+ size_t outLen = 0;
+ for ( size_t n = 0; n < inLen; n++ )
{
wxUint16 cc[2];
+ const size_t numChars = encode_utf16(*inBuff++, cc);
+ if ( numChars == wxCONV_FAILED )
+ return wxCONV_FAILED;
- size_t pa=encode_utf16(*(wxUint32*)psz, cc);
- if (pa == (size_t)-1)
- return pa;
-
- if (buf)
+ outLen += numChars;
+ if ( dst )
{
- *buf++ = cc[0];
- if (pa > 1)
- *buf++ = cc[1];
+ if ( outLen > dstLen )
+ return wxCONV_FAILED;
+
+ *dst++ = cc[0];
+ if ( numChars == 2 )
+ {
+ // second character of a surrogate
+ *dst++ = cc[1];
+ }
}
- len += pa;
- psz += sizeof(wxUint32);
}
- if (buf && len<n) *buf=0;
- return len;
+ return outLen;
}
-
-// copy 16bit String to 32bit MB
-size_t wxMBConvUTF32straight::WC2MB(char *buf, const wchar_t *psz, size_t n) const
+size_t
+wxMBConvUTF32straight::FromWChar(char *dst, size_t dstLen,
+ const wchar_t *src, size_t srcLen) const
{
- size_t len=0;
+ if ( srcLen == wxNO_LEN )
+ srcLen = wxWcslen(src) + 1;
- while (*psz && (!buf || len < n))
+ if ( !dst )
{
- wxUint32 cc;
+ // optimization: return maximal space which could be needed for this
+ // string instead of the exact amount which could be less if there are
+ // any surrogates in the input
+ //
+ // we consider that surrogates are rare enough to make it worthwhile to
+ // avoid running the loop below at the cost of slightly extra memory
+ // consumption
+ return srcLen * BYTES_PER_CHAR;
+ }
- // cast is ok for WC_UTF16
- size_t pa = decode_utf16((const wxUint16 *)psz, cc);
- if (pa == (size_t)-1)
- return pa;
+ wxUint32 *outBuff = wx_reinterpret_cast(wxUint32 *, dst);
+ size_t outLen = 0;
+ for ( const wchar_t * const srcEnd = src + srcLen; src < srcEnd; )
+ {
+ const wxUint32 ch = wxDecodeSurrogate(&src);
+ if ( !src )
+ return wxCONV_FAILED;
- if (buf)
- {
- *(wxUint32*)buf = cc;
- buf += sizeof(wxUint32);
- }
- len += sizeof(wxUint32);
- psz += pa;
- }
+ outLen += BYTES_PER_CHAR;
- if (buf && len<=n-sizeof(wxUint32))
- *(wxUint32*)buf=0;
+ if ( outLen > dstLen )
+ return wxCONV_FAILED;
- return len;
-}
+ *outBuff++ = ch;
+ }
+ return outLen;
+}
+// ----------------------------------------------------------------------------
+// endian-reversing conversions
+// ----------------------------------------------------------------------------
-// swap 32bit MB to 16bit String
-size_t wxMBConvUTF32swap::MB2WC(wchar_t *buf, const char *psz, size_t n) const
+size_t
+wxMBConvUTF32swap::ToWChar(wchar_t *dst, size_t dstLen,
+ const char *src, size_t srcLen) const
{
- size_t len=0;
+ srcLen = GetLength(src, srcLen);
+ if ( srcLen == wxNO_LEN )
+ return wxCONV_FAILED;
- while (*(wxUint32*)psz && (!buf || len < n))
+ const wxUint32 *inBuff = wx_reinterpret_cast(const wxUint32 *, src);
+ const size_t inLen = srcLen / BYTES_PER_CHAR;
+ size_t outLen = 0;
+ for ( size_t n = 0; n < inLen; n++, inBuff++ )
{
- char tmp[4];
- tmp[0] = psz[3]; tmp[1] = psz[2];
- tmp[2] = psz[1]; tmp[3] = psz[0];
-
-
wxUint16 cc[2];
+ const size_t numChars = encode_utf16(wxUINT32_SWAP_ALWAYS(*inBuff), cc);
+ if ( numChars == wxCONV_FAILED )
+ return wxCONV_FAILED;
- size_t pa=encode_utf16(*(wxUint32*)tmp, cc);
- if (pa == (size_t)-1)
- return pa;
-
- if (buf)
+ outLen += numChars;
+ if ( dst )
{
- *buf++ = cc[0];
- if (pa > 1)
- *buf++ = cc[1];
+ if ( outLen > dstLen )
+ return wxCONV_FAILED;
+
+ *dst++ = cc[0];
+ if ( numChars == 2 )
+ {
+ // second character of a surrogate
+ *dst++ = cc[1];
+ }
}
- len += pa;
- psz += sizeof(wxUint32);
}
- if (buf && len<n)
- *buf=0;
-
- return len;
+ return outLen;
}
-
-// swap 16bit String to 32bit MB
-size_t wxMBConvUTF32swap::WC2MB(char *buf, const wchar_t *psz, size_t n) const
+size_t
+wxMBConvUTF32swap::FromWChar(char *dst, size_t dstLen,
+ const wchar_t *src, size_t srcLen) const
{
- size_t len=0;
+ if ( srcLen == wxNO_LEN )
+ srcLen = wxWcslen(src) + 1;
- while (*psz && (!buf || len < n))
+ if ( !dst )
{
- char cc[4];
+ // optimization: return maximal space which could be needed for this
+ // string instead of the exact amount which could be less if there are
+ // any surrogates in the input
+ //
+ // we consider that surrogates are rare enough to make it worthwhile to
+ // avoid running the loop below at the cost of slightly extra memory
+ // consumption
+ return srcLen*BYTES_PER_CHAR;
+ }
- // cast is ok for WC_UTF16
- size_t pa=decode_utf16((const wxUint16 *)psz, *(wxUint32*)cc);
- if (pa == (size_t)-1)
- return pa;
+ wxUint32 *outBuff = wx_reinterpret_cast(wxUint32 *, dst);
+ size_t outLen = 0;
+ for ( const wchar_t * const srcEnd = src + srcLen; src < srcEnd; )
+ {
+ const wxUint32 ch = wxDecodeSurrogate(&src);
+ if ( !src )
+ return wxCONV_FAILED;
- if (buf)
- {
- *buf++ = cc[3];
- *buf++ = cc[2];
- *buf++ = cc[1];
- *buf++ = cc[0];
- }
- len += sizeof(wxUint32);
- psz += pa;
- }
+ outLen += BYTES_PER_CHAR;
- if (buf && len<=n-sizeof(wxUint32))
- *(wxUint32*)buf=0;
+ if ( outLen > dstLen )
+ return wxCONV_FAILED;
- return len;
+ *outBuff++ = wxUINT32_SWAP_ALWAYS(ch);
+ }
+
+ return outLen;
}
-#else // WC_UTF16
+#else // !WC_UTF16: wchar_t is UTF-32
+// ----------------------------------------------------------------------------
+// conversions without endianness change
+// ----------------------------------------------------------------------------
-// copy 32bit MB to 32bit String
-size_t wxMBConvUTF32straight::MB2WC(wchar_t *buf, const char *psz, size_t n) const
+size_t
+wxMBConvUTF32straight::ToWChar(wchar_t *dst, size_t dstLen,
+ const char *src, size_t srcLen) const
{
- size_t len=0;
+ // use memcpy() as it should be much faster than hand-written loop
+ srcLen = GetLength(src, srcLen);
+ if ( srcLen == wxNO_LEN )
+ return wxCONV_FAILED;
- while (*(wxUint32*)psz && (!buf || len < n))
+ const size_t inLen = srcLen/BYTES_PER_CHAR;
+ if ( dst )
{
- if (buf)
- *buf++ = (wchar_t)(*(wxUint32*)psz);
- len++;
- psz += sizeof(wxUint32);
- }
+ if ( dstLen < inLen )
+ return wxCONV_FAILED;
- if (buf && len<n)
- *buf=0;
+ memcpy(dst, src, srcLen);
+ }
- return len;
+ return inLen;
}
-
-// copy 32bit String to 32bit MB
-size_t wxMBConvUTF32straight::WC2MB(char *buf, const wchar_t *psz, size_t n) const
+size_t
+wxMBConvUTF32straight::FromWChar(char *dst, size_t dstLen,
+ const wchar_t *src, size_t srcLen) const
{
- size_t len=0;
+ if ( srcLen == wxNO_LEN )
+ srcLen = wxWcslen(src) + 1;
- while (*psz && (!buf || len < n))
+ srcLen *= BYTES_PER_CHAR;
+
+ if ( dst )
{
- if (buf)
- {
- *(wxUint32*)buf = *psz;
- buf += sizeof(wxUint32);
- }
+ if ( dstLen < srcLen )
+ return wxCONV_FAILED;
- len += sizeof(wxUint32);
- psz++;
+ memcpy(dst, src, srcLen);
}
- if (buf && len<=n-sizeof(wxUint32))
- *(wxUint32*)buf=0;
-
- return len;
+ return srcLen;
}
+// ----------------------------------------------------------------------------
+// endian-reversing conversions
+// ----------------------------------------------------------------------------
-// swap 32bit MB to 32bit String
-size_t wxMBConvUTF32swap::MB2WC(wchar_t *buf, const char *psz, size_t n) const
+size_t
+wxMBConvUTF32swap::ToWChar(wchar_t *dst, size_t dstLen,
+ const char *src, size_t srcLen) const
{
- size_t len=0;
+ srcLen = GetLength(src, srcLen);
+ if ( srcLen == wxNO_LEN )
+ return wxCONV_FAILED;
+
+ srcLen /= BYTES_PER_CHAR;
- while (*(wxUint32*)psz && (!buf || len < n))
+ if ( dst )
{
- if (buf)
+ if ( dstLen < srcLen )
+ return wxCONV_FAILED;
+
+ const wxUint32 *inBuff = wx_reinterpret_cast(const wxUint32 *, src);
+ for ( size_t n = 0; n < srcLen; n++, inBuff++ )
{
- ((char *)buf)[0] = psz[3];
- ((char *)buf)[1] = psz[2];
- ((char *)buf)[2] = psz[1];
- ((char *)buf)[3] = psz[0];
- buf++;
+ *dst++ = wxUINT32_SWAP_ALWAYS(*inBuff);
}
- len++;
- psz += sizeof(wxUint32);
}
- if (buf && len<n)
- *buf=0;
-
- return len;
+ return srcLen;
}
-
-// swap 32bit String to 32bit MB
-size_t wxMBConvUTF32swap::WC2MB(char *buf, const wchar_t *psz, size_t n) const
+size_t
+wxMBConvUTF32swap::FromWChar(char *dst, size_t dstLen,
+ const wchar_t *src, size_t srcLen) const
{
- size_t len=0;
+ if ( srcLen == wxNO_LEN )
+ srcLen = wxWcslen(src) + 1;
- while (*psz && (!buf || len < n))
+ srcLen *= BYTES_PER_CHAR;
+
+ if ( dst )
{
- if (buf)
+ if ( dstLen < srcLen )
+ return wxCONV_FAILED;
+
+ wxUint32 *outBuff = wx_reinterpret_cast(wxUint32 *, dst);
+ for ( size_t n = 0; n < srcLen; n += BYTES_PER_CHAR, src++ )
{
- *buf++ = ((char *)psz)[3];
- *buf++ = ((char *)psz)[2];
- *buf++ = ((char *)psz)[1];
- *buf++ = ((char *)psz)[0];
+ *outBuff++ = wxUINT32_SWAP_ALWAYS(*src);
}
- len += sizeof(wxUint32);
- psz++;
}
- if (buf && len<=n-sizeof(wxUint32))
- *(wxUint32*)buf=0;
-
- return len;
+ return srcLen;
}
-
-#endif // WC_UTF16
+#endif // WC_UTF16/!WC_UTF16
// ============================================================================
class wxMBConv_iconv : public wxMBConv
{
public:
- wxMBConv_iconv(const wxChar *name);
+ wxMBConv_iconv(const char *name);
virtual ~wxMBConv_iconv();
virtual size_t MB2WC(wchar_t *buf, const char *psz, size_t n) const;
virtual size_t WC2MB(char *buf, const wchar_t *psz, size_t n) const;
+ // classify this encoding as explained in wxMBConv::GetMBNulLen() comment
+ virtual size_t GetMBNulLen() const;
+
+#if wxUSE_UNICODE_UTF8
+ virtual bool IsUTF8() const;
+#endif
+
+ virtual wxMBConv *Clone() const
+ {
+ wxMBConv_iconv *p = new wxMBConv_iconv(m_name.ToAscii());
+ p->m_minMBCharWidth = m_minMBCharWidth;
+ return p;
+ }
+
bool IsOk() const
{ return (m2w != ICONV_T_INVALID) && (w2m != ICONV_T_INVALID); }
protected:
- // the iconv handlers used to translate from multibyte to wide char and in
- // the other direction
+ // the iconv handlers used to translate from multibyte
+ // to wide char and in the other direction
iconv_t m2w,
w2m;
+
#if wxUSE_THREADS
// guards access to m2w and w2m objects
wxMutex m_iconvMutex;
// true if the wide char encoding we use (i.e. ms_wcCharsetName) has
// different endian-ness than the native one
static bool ms_wcNeedsSwap;
+
+
+ // name of the encoding handled by this conversion
+ wxString m_name;
+
+ // cached result of GetMBNulLen(); set to 0 meaning "unknown"
+ // initially
+ size_t m_minMBCharWidth;
};
// make the constructor available for unit testing
-WXDLLIMPEXP_BASE wxMBConv* new_wxMBConv_iconv( const wxChar* name )
+WXDLLIMPEXP_BASE wxMBConv* new_wxMBConv_iconv( const char* name )
{
wxMBConv_iconv* result = new wxMBConv_iconv( name );
if ( !result->IsOk() )
delete result;
return 0;
}
+
return result;
}
wxString wxMBConv_iconv::ms_wcCharsetName;
bool wxMBConv_iconv::ms_wcNeedsSwap = false;
-wxMBConv_iconv::wxMBConv_iconv(const wxChar *name)
+wxMBConv_iconv::wxMBConv_iconv(const char *name)
+ : m_name(name)
{
- // iconv operates with chars, not wxChars, but luckily it uses only ASCII
- // names for the charsets
- const wxCharBuffer cname(wxString(name).ToAscii());
+ m_minMBCharWidth = 0;
// check for charset that represents wchar_t:
if ( ms_wcCharsetName.empty() )
#if wxUSE_FONTMAP
const wxChar **names = wxFontMapperBase::GetAllEncodingNames(WC_ENC);
#else // !wxUSE_FONTMAP
- static const wxChar *names[] =
+ static const wxChar *names_static[] =
{
#if SIZEOF_WCHAR_T == 4
_T("UCS-4"),
#endif
NULL
};
+ const wxChar **names = names_static;
#endif // wxUSE_FONTMAP/!wxUSE_FONTMAP
for ( ; *names && ms_wcCharsetName.empty(); ++names )
// first try charset with explicit bytesex info (e.g. "UCS-4LE"):
wxString nameXE(nameCS);
- #ifdef WORDS_BIGENDIAN
+
+#ifdef WORDS_BIGENDIAN
nameXE += _T("BE");
- #else // little endian
+#else // little endian
nameXE += _T("LE");
- #endif
+#endif
wxLogTrace(TRACE_STRCONV, _T(" trying charset \"%s\""),
nameXE.c_str());
- m2w = iconv_open(nameXE.ToAscii(), cname);
+ m2w = iconv_open(nameXE.ToAscii(), name);
if ( m2w == ICONV_T_INVALID )
{
// try charset w/o bytesex info (e.g. "UCS4")
wxLogTrace(TRACE_STRCONV, _T(" trying charset \"%s\""),
nameCS.c_str());
- m2w = iconv_open(nameCS.ToAscii(), cname);
+ m2w = iconv_open(nameCS.ToAscii(), name);
// and check for bytesex ourselves:
if ( m2w != ICONV_T_INVALID )
wbufPtr = wbuf;
bufPtr = buf;
- res = iconv(m2w, ICONV_CHAR_CAST(&bufPtr), &insz,
- (char**)&wbufPtr, &outsz);
+ res = iconv(
+ m2w, ICONV_CHAR_CAST(&bufPtr), &insz,
+ (char**)&wbufPtr, &outsz);
if (ICONV_FAILED(res, insz))
{
wxLogTrace(TRACE_STRCONV,
wxT("iconv wchar_t charset is \"%s\"%s"),
- ms_wcCharsetName.empty() ? _T("<none>")
- : ms_wcCharsetName.c_str(),
+ ms_wcCharsetName.empty() ? wxString("<none>")
+ : ms_wcCharsetName,
ms_wcNeedsSwap ? _T(" (needs swap)")
: _T(""));
}
else // we already have ms_wcCharsetName
{
- m2w = iconv_open(ms_wcCharsetName.ToAscii(), cname);
+ m2w = iconv_open(ms_wcCharsetName.ToAscii(), name);
}
if ( ms_wcCharsetName.empty() )
}
else
{
- w2m = iconv_open(cname, ms_wcCharsetName.ToAscii());
+ w2m = iconv_open(name, ms_wcCharsetName.ToAscii());
if ( w2m == ICONV_T_INVALID )
{
wxLogTrace(TRACE_STRCONV,
wxT("\"%s\" -> \"%s\" works but not the converse!?"),
- ms_wcCharsetName.c_str(), cname.data());
+ ms_wcCharsetName.c_str(), name);
}
}
}
size_t wxMBConv_iconv::MB2WC(wchar_t *buf, const char *psz, size_t n) const
{
+ // find the string length: notice that must be done differently for
+ // NUL-terminated strings and UTF-16/32 which are terminated with 2/4 NULs
+ size_t inbuf;
+ const size_t nulLen = GetMBNulLen();
+ switch ( nulLen )
+ {
+ default:
+ return wxCONV_FAILED;
+
+ case 1:
+ inbuf = strlen(psz); // arguably more optimized than our version
+ break;
+
+ case 2:
+ case 4:
+ // for UTF-16/32 not only we need to have 2/4 consecutive NULs but
+ // they also have to start at character boundary and not span two
+ // adjacent characters
+ const char *p;
+ for ( p = psz; NotAllNULs(p, nulLen); p += nulLen )
+ ;
+ inbuf = p - psz;
+ break;
+ }
+
#if wxUSE_THREADS
- // NB: iconv() is MT-safe, but each thread must use it's own iconv_t handle.
- // Unfortunately there is a couple of global wxCSConv objects such as
+ // NB: iconv() is MT-safe, but each thread must use its own iconv_t handle.
+ // Unfortunately there are a couple of global wxCSConv objects such as
// wxConvLocal that are used all over wx code, so we have to make sure
// the handle is used by at most one thread at the time. Otherwise
// only a few wx classes would be safe to use from non-main threads
// as MB<->WC conversion would fail "randomly".
wxMutexLocker lock(wxConstCast(this, wxMBConv_iconv)->m_iconvMutex);
-#endif
+#endif // wxUSE_THREADS
- size_t inbuf = strlen(psz);
size_t outbuf = n * SIZEOF_WCHAR_T;
size_t res, cres;
// VS: Use these instead of psz, buf because iconv() modifies its arguments:
buf[n] = WC_BSWAP(buf[i]);
}
- // NB: iconv was given only strlen(psz) characters on input, and so
- // it couldn't convert the trailing zero. Let's do it ourselves
- // if there's some room left for it in the output buffer.
+ // NUL-terminate the string if there is any space left
if (res < n)
buf[res] = 0;
}
// to calculate destination buffer requirement
wchar_t tbuf[8];
res = 0;
- do {
+
+ do
+ {
bufPtr = tbuf;
- outbuf = 8*SIZEOF_WCHAR_T;
+ outbuf = 8 * SIZEOF_WCHAR_T;
cres = iconv(m2w,
ICONV_CHAR_CAST(&pszPtr), &inbuf,
(char**)&bufPtr, &outbuf );
- res += 8-(outbuf/SIZEOF_WCHAR_T);
- } while ((cres==(size_t)-1) && (errno==E2BIG));
+ res += 8 - (outbuf / SIZEOF_WCHAR_T);
+ }
+ while ((cres == (size_t)-1) && (errno == E2BIG));
}
if (ICONV_FAILED(cres, inbuf))
{
//VS: it is ok if iconv fails, hence trace only
wxLogTrace(TRACE_STRCONV, wxT("iconv failed: %s"), wxSysErrorMsg(wxSysErrorCode()));
- return (size_t)-1;
+ return wxCONV_FAILED;
}
return res;
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;
}
// have destination buffer, convert there
cres = iconv( w2m, ICONV_CHAR_CAST(&psz), &inbuf, &buf, &outbuf );
- res = n-outbuf;
+ res = n - outbuf;
// NB: iconv was given only wcslen(psz) characters on input, and so
// it couldn't convert the trailing zero. Let's do it ourselves
}
else
{
- // no destination buffer... convert using temp buffer
+ // no destination buffer: convert using temp buffer
// to calculate destination buffer requirement
char tbuf[16];
res = 0;
- do {
- buf = tbuf; outbuf = 16;
+ do
+ {
+ buf = tbuf;
+ outbuf = 16;
cres = iconv( w2m, ICONV_CHAR_CAST(&psz), &inbuf, &buf, &outbuf );
res += 16 - outbuf;
- } while ((cres==(size_t)-1) && (errno==E2BIG));
+ }
+ while ((cres == (size_t)-1) && (errno == E2BIG));
}
if (ms_wcNeedsSwap)
if (ICONV_FAILED(cres, inbuf))
{
wxLogTrace(TRACE_STRCONV, wxT("iconv failed: %s"), wxSysErrorMsg(wxSysErrorCode()));
- return (size_t)-1;
+ return wxCONV_FAILED;
}
return res;
}
+size_t wxMBConv_iconv::GetMBNulLen() const
+{
+ if ( m_minMBCharWidth == 0 )
+ {
+ wxMBConv_iconv * const self = wxConstCast(this, wxMBConv_iconv);
+
+#if wxUSE_THREADS
+ // NB: explained in MB2WC
+ wxMutexLocker lock(self->m_iconvMutex);
+#endif
+
+ const wchar_t *wnul = L"";
+ char buf[8]; // should be enough for NUL in any encoding
+ size_t inLen = sizeof(wchar_t),
+ outLen = WXSIZEOF(buf);
+ char *inBuff = (char *)wnul;
+ char *outBuff = buf;
+ if ( iconv(w2m, ICONV_CHAR_CAST(&inBuff), &inLen, &outBuff, &outLen) == (size_t)-1 )
+ {
+ self->m_minMBCharWidth = (size_t)-1;
+ }
+ else // ok
+ {
+ self->m_minMBCharWidth = outBuff - buf;
+ }
+ }
+
+ return m_minMBCharWidth;
+}
+
+#if wxUSE_UNICODE_UTF8
+bool wxMBConv_iconv::IsUTF8() const
+{
+ return wxStricmp(m_name, "UTF-8") == 0 ||
+ wxStricmp(m_name, "UTF8") == 0;
+}
+#endif
+
#endif // HAVE_ICONV
// from utils.cpp
#if wxUSE_FONTMAP
-extern WXDLLIMPEXP_BASE long wxCharsetToCodepage(const wxChar *charset);
+extern WXDLLIMPEXP_BASE long wxCharsetToCodepage(const char *charset);
extern WXDLLIMPEXP_BASE long wxEncodingToCodepage(wxFontEncoding encoding);
#endif
wxMBConv_win32()
{
m_CodePage = CP_ACP;
+ m_minMBCharWidth = 0;
+ }
+
+ wxMBConv_win32(const wxMBConv_win32& conv)
+ : wxMBConv()
+ {
+ m_CodePage = conv.m_CodePage;
+ m_minMBCharWidth = conv.m_minMBCharWidth;
}
#if wxUSE_FONTMAP
- wxMBConv_win32(const wxChar* name)
+ wxMBConv_win32(const char* name)
{
m_CodePage = wxCharsetToCodepage(name);
+ m_minMBCharWidth = 0;
}
wxMBConv_win32(wxFontEncoding encoding)
{
m_CodePage = wxEncodingToCodepage(encoding);
+ m_minMBCharWidth = 0;
}
-#endif
+#endif // wxUSE_FONTMAP
- size_t MB2WC(wchar_t *buf, const char *psz, size_t n) const
+ virtual size_t MB2WC(wchar_t *buf, const char *psz, size_t n) const
{
// note that we have to use MB_ERR_INVALID_CHARS flag as it without it
// the behaviour is not compatible with the Unix version (using iconv)
// 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:
+ // Win XP or newer and it is not supported for UTF-[78] so we always
+ // use our own conversions in this case. See
// 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() )
+ if ( m_CodePage == CP_UTF8 )
{
- flags = MB_ERR_INVALID_CHARS;
+ return wxMBConvUTF8().MB2WC(buf, psz, n);
+ }
+
+ if ( m_CodePage == CP_UTF7 )
+ {
+ return wxMBConvUTF7().MB2WC(buf, psz, n);
}
- else if ( m_CodePage == CP_UTF8 )
+
+ int flags = 0;
+ if ( (m_CodePage < 50000 && m_CodePage != CP_SYMBOL) &&
+ IsAtLeastWin2kSP4() )
{
- // Avoid round-trip in the special case of UTF-8 by using our
- // own UTF-8 conversion code:
- return wxMBConvUTF8().MB2WC(buf, psz, n);
+ flags = MB_ERR_INVALID_CHARS;
}
const size_t len = ::MultiByteToWideChar
if ( !len )
{
// function totally failed
- return (size_t)-1;
+ return wxCONV_FAILED;
}
// if we were really converting and didn't use MB_ERR_INVALID_CHARS,
{
// we didn't obtain the same thing we started from, hence
// the conversion was lossy and we consider that it failed
- return (size_t)-1;
+ return wxCONV_FAILED;
}
}
return len - 1;
}
- size_t WC2MB(char *buf, const wchar_t *pwz, size_t n) const
+ virtual size_t WC2MB(char *buf, const wchar_t *pwz, size_t n) const
{
/*
we have a problem here: by default, WideCharToMultiByte() may
if ( !len )
{
// function totally failed
- return (size_t)-1;
+ return wxCONV_FAILED;
}
// if we were really converting, check if we succeeded
// check if the conversion failed, i.e. if any replacements
// were done
if ( usedDef )
- return (size_t)-1;
+ return wxCONV_FAILED;
}
else // we must resort to double tripping...
{
wxWCharBuffer wcBuf(n);
- if ( MB2WC(wcBuf.data(), buf, n) == (size_t)-1 ||
+ if ( MB2WC(wcBuf.data(), buf, n) == wxCONV_FAILED ||
wcscmp(wcBuf, pwz) != 0 )
{
// we didn't obtain the same thing we started from, hence
// the conversion was lossy and we consider that it failed
- return (size_t)-1;
+ return wxCONV_FAILED;
}
}
}
- // see the comment above for the reason of "len - 1"
- return len - 1;
+ // see the comment above for the reason of "len - 1"
+ return len - 1;
+ }
+
+ virtual size_t GetMBNulLen() const
+ {
+ if ( m_minMBCharWidth == 0 )
+ {
+ int len = ::WideCharToMultiByte
+ (
+ m_CodePage, // code page
+ 0, // no flags
+ L"", // input string
+ 1, // translate just the NUL
+ NULL, // output buffer
+ 0, // and its size
+ NULL, // no replacement char
+ NULL // [out] don't care if it was used
+ );
+
+ wxMBConv_win32 * const self = wxConstCast(this, wxMBConv_win32);
+ switch ( len )
+ {
+ default:
+ wxLogDebug(_T("Unexpected NUL length %d"), len);
+ self->m_minMBCharWidth = (size_t)-1;
+ break;
+
+ case 0:
+ self->m_minMBCharWidth = (size_t)-1;
+ break;
+
+ case 1:
+ case 2:
+ case 4:
+ self->m_minMBCharWidth = len;
+ break;
+ }
+ }
+
+ return m_minMBCharWidth;
}
+ virtual wxMBConv *Clone() const { return new wxMBConv_win32(*this); }
+
bool IsOk() const { return m_CodePage != -1; }
private:
int verMaj, verMin;
switch ( wxGetOsVersion(&verMaj, &verMin) )
{
- case wxWIN95:
+ case wxOS_WINDOWS_9X:
s_isWin98Or2k = verMaj >= 4 && verMin >= 10;
break;
- case wxWINDOWS_NT:
+ case wxOS_WINDOWS_NT:
s_isWin98Or2k = verMaj >= 5;
break;
default:
- // unknown, be conseravtive by default
+ // unknown: be conservative by default
s_isWin98Or2k = 0;
+ break;
}
wxASSERT_MSG( s_isWin98Or2k != -1, _T("should be set above") );
#endif
}
- 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 <CoreFoundation/CFString.h>
-#include <CoreFoundation/CFStringEncodingExt.h>
-
-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);
- }
+ // the code page we're working with
+ long m_CodePage;
-private:
- CFStringEncoding m_encoding ;
+ // cached result of GetMBNulLen(), set to 0 initially meaning
+ // "unknown"
+ size_t m_minMBCharWidth;
};
-#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 // wxHAVE_WIN32_MB2WC
-#endif // defined(__WXMAC__) && defined(TARGET_CARBON)
// ============================================================================
// wxEncodingConverter based conversion classes
private:
void Init()
{
- m_ok = m2w.Init(m_enc, wxFONTENCODING_UNICODE) &&
+ // Refuse to use broken wxEncodingConverter code for Mac-specific encodings.
+ // The wxMBConv_cf class does a better job.
+ m_ok = (m_enc < wxFONTENCODING_MACMIN || m_enc > wxFONTENCODING_MACMAX) &&
+ 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)
+ wxMBConv_wxwin(const char* name)
{
if (name)
m_enc = wxFontMapperBase::Get()->CharsetToEncoding(name, false);
size_t inbuf = strlen(psz);
if (buf)
{
- if (!m2w.Convert(psz,buf))
- return (size_t)-1;
+ if (!m2w.Convert(psz, buf))
+ return wxCONV_FAILED;
}
return inbuf;
}
const size_t inbuf = wxWcslen(psz);
if (buf)
{
- if (!w2m.Convert(psz,buf))
- return (size_t)-1;
+ if (!w2m.Convert(psz, buf))
+ return wxCONV_FAILED;
}
return inbuf;
}
+ virtual size_t GetMBNulLen() const
+ {
+ switch ( m_enc )
+ {
+ case wxFONTENCODING_UTF16BE:
+ case wxFONTENCODING_UTF16LE:
+ return 2;
+
+ case wxFONTENCODING_UTF32BE:
+ case wxFONTENCODING_UTF32LE:
+ return 4;
+
+ default:
+ return 1;
+ }
+ }
+
+ virtual wxMBConv *Clone() const { return new wxMBConv_wxwin(m_enc); }
+
bool IsOk() const { return m_ok; }
public:
wxFontEncoding m_enc;
wxEncodingConverter m2w, w2m;
+private:
// were we initialized successfully?
bool m_ok;
};
// make the constructors available for unit testing
-WXDLLIMPEXP_BASE wxMBConv* new_wxMBConv_wxwin( const wxChar* name )
+WXDLLIMPEXP_BASE wxMBConv* new_wxMBConv_wxwin( const char* name )
{
wxMBConv_wxwin* result = new wxMBConv_wxwin( name );
if ( !result->IsOk() )
delete result;
return 0;
}
+
return result;
}
m_deferred = true;
}
-wxCSConv::wxCSConv(const wxChar *charset)
+wxCSConv::wxCSConv(const wxString& charset)
{
Init();
- if ( charset )
+ if ( !charset.empty() )
{
- SetName(charset);
+ SetName(charset.ToAscii());
}
#if wxUSE_FONTMAP
m_convReal = NULL;
}
-void wxCSConv::SetName(const wxChar *charset)
+void wxCSConv::SetName(const char *charset)
{
if (charset)
{
}
#if wxUSE_FONTMAP
-#include "wx/hashmap.h"
WX_DECLARE_HASH_MAP( wxFontEncoding, wxString, wxIntegerHash, wxIntegerEqual,
wxEncodingNameCache );
wxLogTrace(TRACE_STRCONV,
wxT("creating conversion for %s"),
(m_name ? m_name
- : wxFontMapperBase::GetEncodingName(m_encoding).c_str()));
+ : (const char*)wxFontMapperBase::GetEncodingName(m_encoding).mb_str()));
#endif // wxUSE_FONTMAP
// check for the special case of ASCII or ISO8859-1 charset: as we have
if ( m_name )
#endif // !wxUSE_FONTMAP
{
- wxString name(m_name);
+#if wxUSE_FONTMAP
wxFontEncoding encoding(m_encoding);
+#endif
- if ( !name.empty() )
+ if ( m_name )
{
- wxMBConv_iconv *conv = new wxMBConv_iconv(name);
+ wxMBConv_iconv *conv = new wxMBConv_iconv(m_name);
if ( conv->IsOk() )
return conv;
#if wxUSE_FONTMAP
encoding =
- wxFontMapperBase::Get()->CharsetToEncoding(name, false);
+ wxFontMapperBase::Get()->CharsetToEncoding(m_name, false);
#endif // wxUSE_FONTMAP
}
#if wxUSE_FONTMAP
if ( it->second.empty() )
return NULL;
- wxMBConv_iconv *conv = new wxMBConv_iconv(it->second);
+ wxMBConv_iconv *conv = new wxMBConv_iconv(it->second.ToAscii());
if ( conv->IsOk() )
return conv;
}
const wxChar** names = wxFontMapperBase::GetAllEncodingNames(encoding);
-
- for ( ; *names; ++names )
+ // CS : in case this does not return valid names (eg for MacRoman)
+ // encoding got a 'failure' entry in the cache all the same,
+ // although it just has to be created using a different method, so
+ // only store failed iconv creation attempts (or perhaps we
+ // shoulnd't do this at all ?)
+ if ( names[0] != NULL )
{
- wxMBConv_iconv *conv = new wxMBConv_iconv(*names);
- if ( conv->IsOk() )
+ for ( ; *names; ++names )
{
- gs_nameCache[encoding] = *names;
- return conv;
+ // FIXME-UTF8: wxFontMapperBase::GetAllEncodingNames()
+ // will need changes that will obsolete this
+ wxString name(*names);
+ wxMBConv_iconv *conv = new wxMBConv_iconv(name.ToAscii());
+ if ( conv->IsOk() )
+ {
+ gs_nameCache[encoding] = *names;
+ return conv;
+ }
+
+ delete conv;
}
- delete conv;
+ gs_nameCache[encoding] = _T(""); // cache the failure
}
-
- gs_nameCache[encoding] = _T(""); // cache the failure
}
#endif // wxUSE_FONTMAP
}
#endif
}
#endif // wxHAVE_WIN32_MB2WC
-#if defined(__WXMAC__)
+
+#ifdef __DARWIN__
{
// 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);
+ wxMBConv_cf *conv = m_name ? new wxMBConv_cf(m_name)
+ : new wxMBConv_cf(m_encoding);
#else
- wxMBConv_mac *conv = new wxMBConv_mac(m_encoding);
-#endif
- if ( conv->IsOk() )
- return conv;
-
- delete conv;
- }
- }
+ wxMBConv_cf *conv = new wxMBConv_cf(m_encoding);
#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
+#endif // __DARWIN__
+
// step (2)
wxFontEncoding enc = m_encoding;
#if wxUSE_FONTMAP
default:
// nothing to do but put here to suppress gcc warnings
- ;
+ break;
}
// step (3)
// NB: This is a hack to prevent deadlock. What could otherwise happen
// in Unicode build: wxConvLocal creation ends up being here
// because of some failure and logs the error. But wxLog will try to
- // attach timestamp, for which it will need wxConvLocal (to convert
- // time to char* and then wchar_t*), but that fails, tries to log
- // error, but wxLog has a (already locked) critical section that
- // guards static buffer.
+ // attach a timestamp, for which it will need wxConvLocal (to convert
+ // time to char* and then wchar_t*), but that fails, tries to log the
+ // error, but wxLog has an (already locked) critical section that
+ // guards the static buffer.
static bool alreadyLoggingError = false;
if (!alreadyLoggingError)
{
m_name ? m_name
:
#if wxUSE_FONTMAP
- wxFontMapperBase::GetEncodingDescription(m_encoding).c_str()
+ (const char*)wxFontMapperBase::GetEncodingDescription(m_encoding).ToAscii()
#else // !wxUSE_FONTMAP
- wxString::Format(_("encoding %s"), m_encoding).c_str()
+ (const char*)wxString::Format(_("encoding %i"), m_encoding).ToAscii()
#endif // wxUSE_FONTMAP/!wxUSE_FONTMAP
);
+
alreadyLoggingError = false;
}
{
wxCSConv *self = (wxCSConv *)this; // const_cast
-#if wxUSE_INTL
// if we don't have neither the name nor the encoding, use the default
// encoding for this system
if ( !m_name && m_encoding == wxFONTENCODING_SYSTEM )
{
- self->m_name = wxStrdup(wxLocale::GetSystemEncodingName());
- }
+#if wxUSE_INTL
+ self->m_encoding = wxLocale::GetSystemEncoding();
+#else
+ // fallback to some reasonable default:
+ self->m_encoding = wxFONTENCODING_ISO8859_1;
#endif // wxUSE_INTL
+ }
self->m_convReal = DoCreate();
self->m_deferred = false;
}
}
+bool wxCSConv::IsOk() const
+{
+ CreateConvIfNeeded();
+
+ // special case: no convReal created for wxFONTENCODING_ISO8859_1
+ if ( m_encoding == wxFONTENCODING_ISO8859_1 )
+ return true; // always ok as we do it ourselves
+
+ // m_convReal->IsOk() is called at its own creation, so we know it must
+ // be ok if m_convReal is non-NULL
+ return m_convReal != NULL;
+}
+
+size_t wxCSConv::ToWChar(wchar_t *dst, size_t dstLen,
+ const char *src, size_t srcLen) const
+{
+ CreateConvIfNeeded();
+
+ if (m_convReal)
+ return m_convReal->ToWChar(dst, dstLen, src, srcLen);
+
+ // latin-1 (direct)
+ return wxMBConv::ToWChar(dst, dstLen, src, srcLen);
+}
+
+size_t wxCSConv::FromWChar(char *dst, size_t dstLen,
+ const wchar_t *src, size_t srcLen) const
+{
+ CreateConvIfNeeded();
+
+ if (m_convReal)
+ return m_convReal->FromWChar(dst, dstLen, src, srcLen);
+
+ // latin-1 (direct)
+ return wxMBConv::FromWChar(dst, dstLen, src, srcLen);
+}
+
size_t wxCSConv::MB2WC(wchar_t *buf, const char *psz, size_t n) const
{
CreateConvIfNeeded();
for (size_t c = 0; c <= len; c++)
{
if (psz[c] > 0xFF)
- return (size_t)-1;
+ return wxCONV_FAILED;
+
buf[c] = (char)psz[c];
}
}
for (size_t c = 0; c <= len; c++)
{
if (psz[c] > 0xFF)
- return (size_t)-1;
+ return wxCONV_FAILED;
}
}
return len;
}
+size_t wxCSConv::GetMBNulLen() const
+{
+ CreateConvIfNeeded();
+
+ if ( m_convReal )
+ {
+ return m_convReal->GetMBNulLen();
+ }
+
+ // otherwise, we are ISO-8859-1
+ return 1;
+}
+
+#if wxUSE_UNICODE_UTF8
+bool wxCSConv::IsUTF8() const
+{
+ CreateConvIfNeeded();
+
+ if ( m_convReal )
+ {
+ return m_convReal->IsUTF8();
+ }
+
+ // otherwise, we are ISO-8859-1
+ return false;
+}
+#endif
+
+
+#if wxUSE_UNICODE
+
+wxWCharBuffer wxSafeConvertMB2WX(const char *s)
+{
+ if ( !s )
+ return wxWCharBuffer();
+
+ wxWCharBuffer wbuf(wxConvLibc.cMB2WX(s));
+ if ( !wbuf )
+ wbuf = wxMBConvUTF8().cMB2WX(s);
+ if ( !wbuf )
+ wbuf = wxConvISO8859_1.cMB2WX(s);
+
+ return wbuf;
+}
+
+wxCharBuffer wxSafeConvertWX2MB(const wchar_t *ws)
+{
+ if ( !ws )
+ return wxCharBuffer();
+
+ wxCharBuffer buf(wxConvLibc.cWX2MB(ws));
+ if ( !buf )
+ buf = wxMBConvUTF8(wxMBConvUTF8::MAP_INVALID_UTF8_TO_OCTAL).cWX2MB(ws);
+
+ return buf;
+}
+
+#endif // wxUSE_UNICODE
+
// ----------------------------------------------------------------------------
// globals
// ----------------------------------------------------------------------------
+// NB: The reason why we create converted objects in this convoluted way,
+// using a factory function instead of global variable, is that they
+// may be used at static initialization time (some of them are used by
+// wxString ctors and there may be a global wxString object). In other
+// words, possibly _before_ the converter global object would be
+// initialized.
+
+#undef wxConvLibc
+#undef wxConvUTF8
+#undef wxConvUTF7
+#undef wxConvLocal
+#undef wxConvISO8859_1
+
+#define WX_DEFINE_GLOBAL_CONV2(klass, impl_klass, name, ctor_args) \
+ WXDLLIMPEXP_DATA_BASE(klass*) name##Ptr = NULL; \
+ WXDLLIMPEXP_BASE klass* wxGet_##name##Ptr() \
+ { \
+ static impl_klass name##Obj ctor_args; \
+ return &name##Obj; \
+ } \
+ /* this ensures that all global converter objects are created */ \
+ /* by the time static initialization is done, i.e. before any */ \
+ /* thread is launched: */ \
+ static klass* gs_##name##instance = wxGet_##name##Ptr()
+
+#define WX_DEFINE_GLOBAL_CONV(klass, name, ctor_args) \
+ WX_DEFINE_GLOBAL_CONV2(klass, klass, name, ctor_args)
+
#ifdef __WINDOWS__
- static wxMBConv_win32 wxConvLibcObj;
-#elif defined(__WXMAC__) && !defined(__MACH__)
- static wxMBConv_mac wxConvLibcObj ;
+ WX_DEFINE_GLOBAL_CONV2(wxMBConv, wxMBConv_win32, wxConvLibc, wxEMPTY_PARAMETER_VALUE);
#else
- static wxMBConvLibc wxConvLibcObj;
+ WX_DEFINE_GLOBAL_CONV2(wxMBConv, wxMBConvLibc, wxConvLibc, wxEMPTY_PARAMETER_VALUE);
#endif
-static wxCSConv wxConvLocalObj(wxFONTENCODING_SYSTEM);
-static wxCSConv wxConvISO8859_1Obj(wxFONTENCODING_ISO8859_1);
-static wxMBConvUTF7 wxConvUTF7Obj;
-static wxMBConvUTF8 wxConvUTF8Obj;
-
-WXDLLIMPEXP_DATA_BASE(wxMBConv&) wxConvLibc = wxConvLibcObj;
-WXDLLIMPEXP_DATA_BASE(wxCSConv&) wxConvLocal = wxConvLocalObj;
-WXDLLIMPEXP_DATA_BASE(wxCSConv&) wxConvISO8859_1 = wxConvISO8859_1Obj;
-WXDLLIMPEXP_DATA_BASE(wxMBConvUTF7&) wxConvUTF7 = wxConvUTF7Obj;
-WXDLLIMPEXP_DATA_BASE(wxMBConvUTF8&) wxConvUTF8 = wxConvUTF8Obj;
-WXDLLIMPEXP_DATA_BASE(wxMBConv *) wxConvCurrent = &wxConvLibcObj;
-WXDLLIMPEXP_DATA_BASE(wxMBConv *) wxConvFileName = &
-#ifdef __WXOSX__
- wxConvUTF8Obj;
-#else
- wxConvLibcObj;
+WX_DEFINE_GLOBAL_CONV(wxMBConvStrictUTF8, wxConvUTF8, wxEMPTY_PARAMETER_VALUE);
+WX_DEFINE_GLOBAL_CONV(wxMBConvUTF7, wxConvUTF7, wxEMPTY_PARAMETER_VALUE);
+
+WX_DEFINE_GLOBAL_CONV(wxCSConv, wxConvLocal, (wxFONTENCODING_SYSTEM));
+WX_DEFINE_GLOBAL_CONV(wxCSConv, wxConvISO8859_1, (wxFONTENCODING_ISO8859_1));
+
+WXDLLIMPEXP_DATA_BASE(wxMBConv *) wxConvCurrent = wxGet_wxConvLibcPtr();
+WXDLLIMPEXP_DATA_BASE(wxMBConv *) wxConvUI = wxGet_wxConvLocalPtr();
+
+#ifdef __DARWIN__
+// The xnu kernel always communicates file paths in decomposed UTF-8.
+// WARNING: Are we sure that CFString's conversion will cause decomposition?
+static wxMBConv_cf wxConvMacUTF8DObj(wxFONTENCODING_UTF8);
#endif
+WXDLLIMPEXP_DATA_BASE(wxMBConv *) wxConvFileName =
+#ifdef __DARWIN__
+ &wxConvMacUTF8DObj;
+#else // !__DARWIN__
+ wxGet_wxConvLibcPtr();
+#endif // __DARWIN__/!__DARWIN__
#else // !wxUSE_WCHAR_T
+// FIXME-UTF8: remove this, wxUSE_WCHAR_T is required now
// stand-ins in absence of wchar_t
WXDLLIMPEXP_DATA_BASE(wxMBConv) wxConvLibc,
wxConvISO8859_1,