1 /////////////////////////////////////////////////////////////////////////////
2 // Name: src/common/strconv.cpp
3 // Purpose: Unicode conversion classes
4 // Author: Ove Kaaven, Robert Roebling, Vadim Zeitlin, Vaclav Slavik,
5 // Ryan Norton, Fredrik Roubert (UTF7)
9 // Copyright: (c) 1999 Ove Kaaven, Robert Roebling, Vaclav Slavik
10 // (c) 2000-2003 Vadim Zeitlin
11 // (c) 2004 Ryan Norton, Fredrik Roubert
12 // Licence: wxWindows licence
13 /////////////////////////////////////////////////////////////////////////////
15 // ============================================================================
17 // ============================================================================
19 // ----------------------------------------------------------------------------
21 // ----------------------------------------------------------------------------
23 // For compilers that support precompilation, includes "wx.h".
24 #include "wx/wxprec.h"
35 #include "wx/strconv.h"
40 #include "wx/msw/private.h"
41 #include "wx/msw/missing.h"
52 #if defined(__WIN32__) && !defined(__WXMICROWIN__)
53 #define wxHAVE_WIN32_MB2WC
54 #endif // __WIN32__ but !__WXMICROWIN__
62 #include "wx/thread.h"
65 #include "wx/encconv.h"
66 #include "wx/fontmap.h"
71 #include <ATSUnicode.h>
72 #include <TextCommon.h>
73 #include <TextEncodingConverter.h>
76 #include "wx/mac/private.h" // includes mac headers
79 #define TRACE_STRCONV _T("strconv")
81 #if SIZEOF_WCHAR_T == 2
85 // ============================================================================
87 // ============================================================================
89 // helper function of cMB2WC(): check if n bytes at this location are all NUL
90 static bool NotAllNULs(const char *p
, size_t n
)
92 while ( n
&& *p
++ == '\0' )
98 // ----------------------------------------------------------------------------
99 // UTF-16 en/decoding to/from UCS-4
100 // ----------------------------------------------------------------------------
103 static size_t encode_utf16(wxUint32 input
, wxUint16
*output
)
108 *output
= (wxUint16
) input
;
111 else if (input
>=0x110000)
119 *output
++ = (wxUint16
) ((input
>> 10)+0xd7c0);
120 *output
= (wxUint16
) ((input
&0x3ff)+0xdc00);
126 static size_t decode_utf16(const wxUint16
* input
, wxUint32
& output
)
128 if ((*input
<0xd800) || (*input
>0xdfff))
133 else if ((input
[1]<0xdc00) || (input
[1]>0xdfff))
140 output
= ((input
[0] - 0xd7c0) << 10) + (input
[1] - 0xdc00);
146 // ----------------------------------------------------------------------------
148 // ----------------------------------------------------------------------------
151 wxMBConv::ToWChar(wchar_t *dst
, size_t dstLen
,
152 const char *src
, size_t srcLen
) const
154 // although new conversion classes are supposed to implement this function
155 // directly, the existins ones only implement the old MB2WC() and so, to
156 // avoid to have to rewrite all conversion classes at once, we provide a
157 // default (but not efficient) implementation of this one in terms of the
158 // old function by copying the input to ensure that it's NUL-terminated and
159 // then using MB2WC() to convert it
161 // the number of chars [which would be] written to dst [if it were not NULL]
162 size_t dstWritten
= 0;
164 // the number of NULs terminating this string
165 size_t nulLen
wxDUMMY_INITIALIZE(0);
167 // if we were not given the input size we just have to assume that the
168 // string is properly terminated as we have no way of knowing how long it
169 // is anyhow, but if we do have the size check whether there are enough
173 if ( srcLen
!= (size_t)-1 )
175 // we need to know how to find the end of this string
176 nulLen
= GetMBNulLen();
177 if ( nulLen
== wxCONV_FAILED
)
178 return wxCONV_FAILED
;
180 // if there are enough NULs we can avoid the copy
181 if ( srcLen
< nulLen
|| NotAllNULs(src
+ srcLen
- nulLen
, nulLen
) )
183 // make a copy in order to properly NUL-terminate the string
184 bufTmp
= wxCharBuffer(srcLen
+ nulLen
- 1 /* 1 will be added */);
185 char * const p
= bufTmp
.data();
186 memcpy(p
, src
, srcLen
);
187 for ( char *s
= p
+ srcLen
; s
< p
+ srcLen
+ nulLen
; s
++ )
193 srcEnd
= src
+ srcLen
;
195 else // quit after the first loop iteration
202 // try to convert the current chunk
203 size_t lenChunk
= MB2WC(NULL
, src
, 0);
206 // nothing left in the input string, conversion succeeded; but
207 // still account for the trailing NULL
212 if ( lenChunk
== wxCONV_FAILED
)
213 return wxCONV_FAILED
;
215 lenChunk
++; // for trailing NUL
217 dstWritten
+= lenChunk
;
221 if ( dstWritten
> dstLen
)
222 return wxCONV_FAILED
;
224 if ( MB2WC(dst
, src
, lenChunk
) == wxCONV_FAILED
)
225 return wxCONV_FAILED
;
232 // we convert the entire string in this cas, as we suppose that the
233 // string is NUL-terminated and so srcEnd is not used at all
237 // advance the input pointer past the end of this chunk
238 while ( NotAllNULs(src
, nulLen
) )
240 // notice that we must skip over multiple bytes here as we suppose
241 // that if NUL takes 2 or 4 bytes, then all the other characters do
242 // too and so if advanced by a single byte we might erroneously
243 // detect sequences of NUL bytes in the middle of the input
247 src
+= nulLen
; // skipping over its terminator as well
249 // note that ">=" (and not just "==") is needed here as the terminator
250 // we skipped just above could be inside or just after the buffer
251 // delimited by inEnd
260 wxMBConv::FromWChar(char *dst
, size_t dstLen
,
261 const wchar_t *src
, size_t srcLen
) const
263 // the number of chars [which would be] written to dst [if it were not NULL]
264 size_t dstWritten
= 0;
266 // make a copy of the input string unless it is already properly
269 // if we don't know its length we have no choice but to assume that it is,
270 // indeed, properly terminated
271 wxWCharBuffer bufTmp
;
272 if ( srcLen
== (size_t)-1 )
274 srcLen
= wxWcslen(src
) + 1;
276 else if ( srcLen
!= 0 && src
[srcLen
- 1] != L
'\0' )
278 // make a copy in order to properly NUL-terminate the string
279 bufTmp
= wxWCharBuffer(srcLen
);
280 memcpy(bufTmp
.data(), src
, srcLen
*sizeof(wchar_t));
284 const size_t lenNul
= GetMBNulLen();
285 for ( const wchar_t * const srcEnd
= src
+ srcLen
;
287 src
+= wxWcslen(src
) + 1 /* skip L'\0' too */ )
289 // try to convert the current chunk
290 size_t lenChunk
= WC2MB(NULL
, src
, 0);
292 if ( lenChunk
== wxCONV_FAILED
)
293 return wxCONV_FAILED
;
296 dstWritten
+= lenChunk
;
300 if ( dstWritten
> dstLen
)
301 return wxCONV_FAILED
;
303 if ( WC2MB(dst
, src
, lenChunk
) == wxCONV_FAILED
)
304 return wxCONV_FAILED
;
313 size_t wxMBConv::MB2WC(wchar_t *out
, const char *in
, size_t outLen
) const
315 size_t rc
= ToWChar(out
, outLen
, in
);
316 if ( rc
!= wxCONV_FAILED
)
318 // ToWChar() returns the buffer length, i.e. including the trailing
319 // NUL, while this method doesn't take it into account
326 size_t wxMBConv::WC2MB(char *out
, const wchar_t *in
, size_t outLen
) const
328 size_t rc
= FromWChar(out
, outLen
, in
);
329 if ( rc
!= wxCONV_FAILED
)
337 wxMBConv::~wxMBConv()
339 // nothing to do here (necessary for Darwin linking probably)
342 const wxWCharBuffer
wxMBConv::cMB2WC(const char *psz
) const
346 // calculate the length of the buffer needed first
347 const size_t nLen
= MB2WC(NULL
, psz
, 0);
348 if ( nLen
!= wxCONV_FAILED
)
350 // now do the actual conversion
351 wxWCharBuffer
buf(nLen
/* +1 added implicitly */);
353 // +1 for the trailing NULL
354 if ( MB2WC(buf
.data(), psz
, nLen
+ 1) != wxCONV_FAILED
)
359 return wxWCharBuffer();
362 const wxCharBuffer
wxMBConv::cWC2MB(const wchar_t *pwz
) const
366 const size_t nLen
= WC2MB(NULL
, pwz
, 0);
367 if ( nLen
!= wxCONV_FAILED
)
369 // extra space for trailing NUL(s)
370 static const size_t extraLen
= GetMaxMBNulLen();
372 wxCharBuffer
buf(nLen
+ extraLen
- 1);
373 if ( WC2MB(buf
.data(), pwz
, nLen
+ extraLen
) != wxCONV_FAILED
)
378 return wxCharBuffer();
382 wxMBConv::cMB2WC(const char *in
, size_t inLen
, size_t *outLen
) const
384 const size_t dstLen
= ToWChar(NULL
, 0, in
, inLen
);
385 if ( dstLen
!= wxCONV_FAILED
)
387 wxWCharBuffer
wbuf(dstLen
- 1);
388 if ( ToWChar(wbuf
.data(), dstLen
, in
, inLen
) )
391 *outLen
= dstLen
- 1;
399 return wxWCharBuffer();
403 wxMBConv::cWC2MB(const wchar_t *in
, size_t inLen
, size_t *outLen
) const
405 const size_t dstLen
= FromWChar(NULL
, 0, in
, inLen
);
406 if ( dstLen
!= wxCONV_FAILED
)
408 wxCharBuffer
buf(dstLen
- 1);
409 if ( FromWChar(buf
.data(), dstLen
, in
, inLen
) )
412 *outLen
= dstLen
- 1;
420 return wxCharBuffer();
423 // ----------------------------------------------------------------------------
425 // ----------------------------------------------------------------------------
427 size_t wxMBConvLibc::MB2WC(wchar_t *buf
, const char *psz
, size_t n
) const
429 return wxMB2WC(buf
, psz
, n
);
432 size_t wxMBConvLibc::WC2MB(char *buf
, const wchar_t *psz
, size_t n
) const
434 return wxWC2MB(buf
, psz
, n
);
437 // ----------------------------------------------------------------------------
438 // wxConvBrokenFileNames
439 // ----------------------------------------------------------------------------
443 wxConvBrokenFileNames::wxConvBrokenFileNames(const wxChar
*charset
)
445 if ( !charset
|| wxStricmp(charset
, _T("UTF-8")) == 0
446 || wxStricmp(charset
, _T("UTF8")) == 0 )
447 m_conv
= new wxMBConvUTF8(wxMBConvUTF8::MAP_INVALID_UTF8_TO_OCTAL
);
449 m_conv
= new wxCSConv(charset
);
454 // ----------------------------------------------------------------------------
456 // ----------------------------------------------------------------------------
458 // Implementation (C) 2004 Fredrik Roubert
461 // BASE64 decoding table
463 static const unsigned char utf7unb64
[] =
465 0xff, 0xff, 0xff, 0xff, 0xff, 0xff, 0xff, 0xff,
466 0xff, 0xff, 0xff, 0xff, 0xff, 0xff, 0xff, 0xff,
467 0xff, 0xff, 0xff, 0xff, 0xff, 0xff, 0xff, 0xff,
468 0xff, 0xff, 0xff, 0xff, 0xff, 0xff, 0xff, 0xff,
469 0xff, 0xff, 0xff, 0xff, 0xff, 0xff, 0xff, 0xff,
470 0xff, 0xff, 0xff, 0x3e, 0xff, 0xff, 0xff, 0x3f,
471 0x34, 0x35, 0x36, 0x37, 0x38, 0x39, 0x3a, 0x3b,
472 0x3c, 0x3d, 0xff, 0xff, 0xff, 0xff, 0xff, 0xff,
473 0xff, 0x00, 0x01, 0x02, 0x03, 0x04, 0x05, 0x06,
474 0x07, 0x08, 0x09, 0x0a, 0x0b, 0x0c, 0x0d, 0x0e,
475 0x0f, 0x10, 0x11, 0x12, 0x13, 0x14, 0x15, 0x16,
476 0x17, 0x18, 0x19, 0xff, 0xff, 0xff, 0xff, 0xff,
477 0xff, 0x1a, 0x1b, 0x1c, 0x1d, 0x1e, 0x1f, 0x20,
478 0x21, 0x22, 0x23, 0x24, 0x25, 0x26, 0x27, 0x28,
479 0x29, 0x2a, 0x2b, 0x2c, 0x2d, 0x2e, 0x2f, 0x30,
480 0x31, 0x32, 0x33, 0xff, 0xff, 0xff, 0xff, 0xff,
481 0xff, 0xff, 0xff, 0xff, 0xff, 0xff, 0xff, 0xff,
482 0xff, 0xff, 0xff, 0xff, 0xff, 0xff, 0xff, 0xff,
483 0xff, 0xff, 0xff, 0xff, 0xff, 0xff, 0xff, 0xff,
484 0xff, 0xff, 0xff, 0xff, 0xff, 0xff, 0xff, 0xff,
485 0xff, 0xff, 0xff, 0xff, 0xff, 0xff, 0xff, 0xff,
486 0xff, 0xff, 0xff, 0xff, 0xff, 0xff, 0xff, 0xff,
487 0xff, 0xff, 0xff, 0xff, 0xff, 0xff, 0xff, 0xff,
488 0xff, 0xff, 0xff, 0xff, 0xff, 0xff, 0xff, 0xff,
489 0xff, 0xff, 0xff, 0xff, 0xff, 0xff, 0xff, 0xff,
490 0xff, 0xff, 0xff, 0xff, 0xff, 0xff, 0xff, 0xff,
491 0xff, 0xff, 0xff, 0xff, 0xff, 0xff, 0xff, 0xff,
492 0xff, 0xff, 0xff, 0xff, 0xff, 0xff, 0xff, 0xff,
493 0xff, 0xff, 0xff, 0xff, 0xff, 0xff, 0xff, 0xff,
494 0xff, 0xff, 0xff, 0xff, 0xff, 0xff, 0xff, 0xff,
495 0xff, 0xff, 0xff, 0xff, 0xff, 0xff, 0xff, 0xff,
496 0xff, 0xff, 0xff, 0xff, 0xff, 0xff, 0xff, 0xff
499 size_t wxMBConvUTF7::MB2WC(wchar_t *buf
, const char *psz
, size_t n
) const
503 while ( *psz
&& (!buf
|| (len
< n
)) )
505 unsigned char cc
= *psz
++;
513 else if (*psz
== '-')
521 else // start of BASE64 encoded string
525 for ( ok
= lsb
= false, d
= 0, l
= 0;
526 (cc
= utf7unb64
[(unsigned char)*psz
]) != 0xff;
531 for (l
+= 6; l
>= 8; lsb
= !lsb
)
533 unsigned char c
= (unsigned char)((d
>> (l
-= 8)) % 256);
543 *buf
= (wchar_t)(c
<< 8);
552 // in valid UTF7 we should have valid characters after '+'
561 if ( buf
&& (len
< n
) )
568 // BASE64 encoding table
570 static const unsigned char utf7enb64
[] =
572 'A', 'B', 'C', 'D', 'E', 'F', 'G', 'H',
573 'I', 'J', 'K', 'L', 'M', 'N', 'O', 'P',
574 'Q', 'R', 'S', 'T', 'U', 'V', 'W', 'X',
575 'Y', 'Z', 'a', 'b', 'c', 'd', 'e', 'f',
576 'g', 'h', 'i', 'j', 'k', 'l', 'm', 'n',
577 'o', 'p', 'q', 'r', 's', 't', 'u', 'v',
578 'w', 'x', 'y', 'z', '0', '1', '2', '3',
579 '4', '5', '6', '7', '8', '9', '+', '/'
583 // UTF-7 encoding table
585 // 0 - Set D (directly encoded characters)
586 // 1 - Set O (optional direct characters)
587 // 2 - whitespace characters (optional)
588 // 3 - special characters
590 static const unsigned char utf7encode
[128] =
592 3, 3, 3, 3, 3, 3, 3, 3, 3, 2, 2, 3, 3, 2, 3, 3,
593 3, 3, 3, 3, 3, 3, 3, 3, 3, 3, 3, 3, 3, 3, 3, 3,
594 2, 1, 1, 1, 1, 1, 1, 0, 0, 0, 1, 3, 0, 0, 0, 3,
595 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 1, 1, 1, 1, 0,
596 1, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0,
597 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 1, 3, 1, 1, 1,
598 1, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0,
599 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 1, 1, 1, 3, 3
602 size_t wxMBConvUTF7::WC2MB(char *buf
, const wchar_t *psz
, size_t n
) const
606 while (*psz
&& ((!buf
) || (len
< n
)))
609 if (cc
< 0x80 && utf7encode
[cc
] < 1)
617 else if (((wxUint32
)cc
) > 0xffff)
619 // no surrogate pair generation (yet?)
630 // BASE64 encode string
631 unsigned int lsb
, d
, l
;
632 for (d
= 0, l
= 0; /*nothing*/; psz
++)
634 for (lsb
= 0; lsb
< 2; lsb
++)
637 d
+= lsb
? cc
& 0xff : (cc
& 0xff00) >> 8;
639 for (l
+= 8; l
>= 6; )
643 *buf
++ = utf7enb64
[(d
>> l
) % 64];
648 if (!(cc
) || (cc
< 0x80 && utf7encode
[cc
] < 1))
654 *buf
++ = utf7enb64
[((d
% 16) << (6 - l
)) % 64];
663 if (buf
&& (len
< n
))
668 // ----------------------------------------------------------------------------
670 // ----------------------------------------------------------------------------
672 static wxUint32 utf8_max
[]=
673 { 0x7f, 0x7ff, 0xffff, 0x1fffff, 0x3ffffff, 0x7fffffff, 0xffffffff };
675 // boundaries of the private use area we use to (temporarily) remap invalid
676 // characters invalid in a UTF-8 encoded string
677 const wxUint32 wxUnicodePUA
= 0x100000;
678 const wxUint32 wxUnicodePUAEnd
= wxUnicodePUA
+ 256;
680 size_t wxMBConvUTF8::MB2WC(wchar_t *buf
, const char *psz
, size_t n
) const
684 while (*psz
&& ((!buf
) || (len
< n
)))
686 const char *opsz
= psz
;
687 bool invalid
= false;
688 unsigned char cc
= *psz
++, fc
= cc
;
690 for (cnt
= 0; fc
& 0x80; cnt
++)
699 // escape the escape character for octal escapes
700 if ((m_options
& MAP_INVALID_UTF8_TO_OCTAL
)
701 && cc
== '\\' && (!buf
|| len
< n
))
713 // invalid UTF-8 sequence
718 unsigned ocnt
= cnt
- 1;
719 wxUint32 res
= cc
& (0x3f >> cnt
);
723 if ((cc
& 0xC0) != 0x80)
725 // invalid UTF-8 sequence
730 res
= (res
<< 6) | (cc
& 0x3f);
732 if (invalid
|| res
<= utf8_max
[ocnt
])
734 // illegal UTF-8 encoding
737 else if ((m_options
& MAP_INVALID_UTF8_TO_PUA
) &&
738 res
>= wxUnicodePUA
&& res
< wxUnicodePUAEnd
)
740 // if one of our PUA characters turns up externally
741 // it must also be treated as an illegal sequence
742 // (a bit like you have to escape an escape character)
748 // cast is ok because wchar_t == wxUuint16 if WC_UTF16
749 size_t pa
= encode_utf16(res
, (wxUint16
*)buf
);
750 if (pa
== (size_t)-1)
762 *buf
++ = (wchar_t)res
;
764 #endif // WC_UTF16/!WC_UTF16
769 if (m_options
& MAP_INVALID_UTF8_TO_PUA
)
771 while (opsz
< psz
&& (!buf
|| len
< n
))
774 // cast is ok because wchar_t == wxUuint16 if WC_UTF16
775 size_t pa
= encode_utf16((unsigned char)*opsz
+ wxUnicodePUA
, (wxUint16
*)buf
);
776 wxASSERT(pa
!= (size_t)-1);
783 *buf
++ = (wchar_t)(wxUnicodePUA
+ (unsigned char)*opsz
);
789 else if (m_options
& MAP_INVALID_UTF8_TO_OCTAL
)
791 while (opsz
< psz
&& (!buf
|| len
< n
))
793 if ( buf
&& len
+ 3 < n
)
795 unsigned char on
= *opsz
;
797 *buf
++ = (wchar_t)( L
'0' + on
/ 0100 );
798 *buf
++ = (wchar_t)( L
'0' + (on
% 0100) / 010 );
799 *buf
++ = (wchar_t)( L
'0' + on
% 010 );
805 else // MAP_INVALID_UTF8_NOT
812 if (buf
&& (len
< n
))
817 static inline bool isoctal(wchar_t wch
)
819 return L
'0' <= wch
&& wch
<= L
'7';
822 size_t wxMBConvUTF8::WC2MB(char *buf
, const wchar_t *psz
, size_t n
) const
826 while (*psz
&& ((!buf
) || (len
< n
)))
830 // cast is ok for WC_UTF16
831 size_t pa
= decode_utf16((const wxUint16
*)psz
, cc
);
832 psz
+= (pa
== (size_t)-1) ? 1 : pa
;
834 cc
=(*psz
++) & 0x7fffffff;
837 if ( (m_options
& MAP_INVALID_UTF8_TO_PUA
)
838 && cc
>= wxUnicodePUA
&& cc
< wxUnicodePUAEnd
)
841 *buf
++ = (char)(cc
- wxUnicodePUA
);
844 else if ( (m_options
& MAP_INVALID_UTF8_TO_OCTAL
)
845 && cc
== L
'\\' && psz
[0] == L
'\\' )
852 else if ( (m_options
& MAP_INVALID_UTF8_TO_OCTAL
) &&
854 isoctal(psz
[0]) && isoctal(psz
[1]) && isoctal(psz
[2]) )
858 *buf
++ = (char) ((psz
[0] - L
'0')*0100 +
859 (psz
[1] - L
'0')*010 +
869 for (cnt
= 0; cc
> utf8_max
[cnt
]; cnt
++) {}
883 *buf
++ = (char) ((-128 >> cnt
) | ((cc
>> (cnt
* 6)) & (0x3f >> cnt
)));
885 *buf
++ = (char) (0x80 | ((cc
>> (cnt
* 6)) & 0x3f));
897 // ----------------------------------------------------------------------------
899 // ----------------------------------------------------------------------------
901 #ifdef WORDS_BIGENDIAN
902 #define wxMBConvUTF16straight wxMBConvUTF16BE
903 #define wxMBConvUTF16swap wxMBConvUTF16LE
905 #define wxMBConvUTF16swap wxMBConvUTF16BE
906 #define wxMBConvUTF16straight wxMBConvUTF16LE
912 // copy 16bit MB to 16bit String
913 size_t wxMBConvUTF16straight::MB2WC(wchar_t *buf
, const char *psz
, size_t n
) const
917 while (*(wxUint16
*)psz
&& (!buf
|| len
< n
))
920 *buf
++ = *(wxUint16
*)psz
;
923 psz
+= sizeof(wxUint16
);
925 if (buf
&& len
<n
) *buf
=0;
931 // copy 16bit String to 16bit MB
932 size_t wxMBConvUTF16straight::WC2MB(char *buf
, const wchar_t *psz
, size_t n
) const
936 while (*psz
&& (!buf
|| len
< n
))
940 *(wxUint16
*)buf
= *psz
;
941 buf
+= sizeof(wxUint16
);
943 len
+= sizeof(wxUint16
);
946 if (buf
&& len
<=n
-sizeof(wxUint16
)) *(wxUint16
*)buf
=0;
952 // swap 16bit MB to 16bit String
953 size_t wxMBConvUTF16swap::MB2WC(wchar_t *buf
, const char *psz
, size_t n
) const
957 // UTF16 string must be terminated by 2 NULs as single NULs may occur
959 while ( (psz
[0] || psz
[1]) && (!buf
|| len
< n
) )
963 ((char *)buf
)[0] = psz
[1];
964 ((char *)buf
)[1] = psz
[0];
971 if ( buf
&& len
< n
)
978 // swap 16bit MB to 16bit String
979 size_t wxMBConvUTF16swap::WC2MB(char *buf
, const wchar_t *psz
, size_t n
) const
983 while ( *psz
&& (!buf
|| len
< n
) )
987 *buf
++ = ((char*)psz
)[1];
988 *buf
++ = ((char*)psz
)[0];
994 if ( buf
&& len
< n
)
1004 // copy 16bit MB to 32bit String
1005 size_t wxMBConvUTF16straight::MB2WC(wchar_t *buf
, const char *psz
, size_t n
) const
1009 while (*(wxUint16
*)psz
&& (!buf
|| len
< n
))
1012 size_t pa
=decode_utf16((wxUint16
*)psz
, cc
);
1013 if (pa
== (size_t)-1)
1017 *buf
++ = (wchar_t)cc
;
1019 psz
+= pa
* sizeof(wxUint16
);
1021 if (buf
&& len
<n
) *buf
=0;
1027 // copy 32bit String to 16bit MB
1028 size_t wxMBConvUTF16straight::WC2MB(char *buf
, const wchar_t *psz
, size_t n
) const
1032 while (*psz
&& (!buf
|| len
< n
))
1035 size_t pa
=encode_utf16(*psz
, cc
);
1037 if (pa
== (size_t)-1)
1042 *(wxUint16
*)buf
= cc
[0];
1043 buf
+= sizeof(wxUint16
);
1046 *(wxUint16
*)buf
= cc
[1];
1047 buf
+= sizeof(wxUint16
);
1051 len
+= pa
*sizeof(wxUint16
);
1054 if (buf
&& len
<=n
-sizeof(wxUint16
)) *(wxUint16
*)buf
=0;
1060 // swap 16bit MB to 32bit String
1061 size_t wxMBConvUTF16swap::MB2WC(wchar_t *buf
, const char *psz
, size_t n
) const
1065 while (*(wxUint16
*)psz
&& (!buf
|| len
< n
))
1069 tmp
[0]=psz
[1]; tmp
[1]=psz
[0];
1070 tmp
[2]=psz
[3]; tmp
[3]=psz
[2];
1072 size_t pa
=decode_utf16((wxUint16
*)tmp
, cc
);
1073 if (pa
== (size_t)-1)
1077 *buf
++ = (wchar_t)cc
;
1080 psz
+= pa
* sizeof(wxUint16
);
1082 if (buf
&& len
<n
) *buf
=0;
1088 // swap 32bit String to 16bit MB
1089 size_t wxMBConvUTF16swap::WC2MB(char *buf
, const wchar_t *psz
, size_t n
) const
1093 while (*psz
&& (!buf
|| len
< n
))
1096 size_t pa
=encode_utf16(*psz
, cc
);
1098 if (pa
== (size_t)-1)
1103 *buf
++ = ((char*)cc
)[1];
1104 *buf
++ = ((char*)cc
)[0];
1107 *buf
++ = ((char*)cc
)[3];
1108 *buf
++ = ((char*)cc
)[2];
1112 len
+= pa
*sizeof(wxUint16
);
1115 if (buf
&& len
<=n
-sizeof(wxUint16
)) *(wxUint16
*)buf
=0;
1123 // ----------------------------------------------------------------------------
1125 // ----------------------------------------------------------------------------
1127 #ifdef WORDS_BIGENDIAN
1128 #define wxMBConvUTF32straight wxMBConvUTF32BE
1129 #define wxMBConvUTF32swap wxMBConvUTF32LE
1131 #define wxMBConvUTF32swap wxMBConvUTF32BE
1132 #define wxMBConvUTF32straight wxMBConvUTF32LE
1136 WXDLLIMPEXP_DATA_BASE(wxMBConvUTF32LE
) wxConvUTF32LE
;
1137 WXDLLIMPEXP_DATA_BASE(wxMBConvUTF32BE
) wxConvUTF32BE
;
1142 // copy 32bit MB to 16bit String
1143 size_t wxMBConvUTF32straight::MB2WC(wchar_t *buf
, const char *psz
, size_t n
) const
1147 while (*(wxUint32
*)psz
&& (!buf
|| len
< n
))
1151 size_t pa
=encode_utf16(*(wxUint32
*)psz
, cc
);
1152 if (pa
== (size_t)-1)
1162 psz
+= sizeof(wxUint32
);
1164 if (buf
&& len
<n
) *buf
=0;
1170 // copy 16bit String to 32bit MB
1171 size_t wxMBConvUTF32straight::WC2MB(char *buf
, const wchar_t *psz
, size_t n
) const
1175 while (*psz
&& (!buf
|| len
< n
))
1179 // cast is ok for WC_UTF16
1180 size_t pa
= decode_utf16((const wxUint16
*)psz
, cc
);
1181 if (pa
== (size_t)-1)
1186 *(wxUint32
*)buf
= cc
;
1187 buf
+= sizeof(wxUint32
);
1189 len
+= sizeof(wxUint32
);
1193 if (buf
&& len
<=n
-sizeof(wxUint32
))
1201 // swap 32bit MB to 16bit String
1202 size_t wxMBConvUTF32swap::MB2WC(wchar_t *buf
, const char *psz
, size_t n
) const
1206 while (*(wxUint32
*)psz
&& (!buf
|| len
< n
))
1209 tmp
[0] = psz
[3]; tmp
[1] = psz
[2];
1210 tmp
[2] = psz
[1]; tmp
[3] = psz
[0];
1215 size_t pa
=encode_utf16(*(wxUint32
*)tmp
, cc
);
1216 if (pa
== (size_t)-1)
1226 psz
+= sizeof(wxUint32
);
1236 // swap 16bit String to 32bit MB
1237 size_t wxMBConvUTF32swap::WC2MB(char *buf
, const wchar_t *psz
, size_t n
) const
1241 while (*psz
&& (!buf
|| len
< n
))
1245 // cast is ok for WC_UTF16
1246 size_t pa
=decode_utf16((const wxUint16
*)psz
, *(wxUint32
*)cc
);
1247 if (pa
== (size_t)-1)
1257 len
+= sizeof(wxUint32
);
1261 if (buf
&& len
<=n
-sizeof(wxUint32
))
1270 // copy 32bit MB to 32bit String
1271 size_t wxMBConvUTF32straight::MB2WC(wchar_t *buf
, const char *psz
, size_t n
) const
1275 while (*(wxUint32
*)psz
&& (!buf
|| len
< n
))
1278 *buf
++ = (wchar_t)(*(wxUint32
*)psz
);
1280 psz
+= sizeof(wxUint32
);
1290 // copy 32bit String to 32bit MB
1291 size_t wxMBConvUTF32straight::WC2MB(char *buf
, const wchar_t *psz
, size_t n
) const
1295 while (*psz
&& (!buf
|| len
< n
))
1299 *(wxUint32
*)buf
= *psz
;
1300 buf
+= sizeof(wxUint32
);
1303 len
+= sizeof(wxUint32
);
1307 if (buf
&& len
<=n
-sizeof(wxUint32
))
1314 // swap 32bit MB to 32bit String
1315 size_t wxMBConvUTF32swap::MB2WC(wchar_t *buf
, const char *psz
, size_t n
) const
1319 while (*(wxUint32
*)psz
&& (!buf
|| len
< n
))
1323 ((char *)buf
)[0] = psz
[3];
1324 ((char *)buf
)[1] = psz
[2];
1325 ((char *)buf
)[2] = psz
[1];
1326 ((char *)buf
)[3] = psz
[0];
1330 psz
+= sizeof(wxUint32
);
1340 // swap 32bit String to 32bit MB
1341 size_t wxMBConvUTF32swap::WC2MB(char *buf
, const wchar_t *psz
, size_t n
) const
1345 while (*psz
&& (!buf
|| len
< n
))
1349 *buf
++ = ((char *)psz
)[3];
1350 *buf
++ = ((char *)psz
)[2];
1351 *buf
++ = ((char *)psz
)[1];
1352 *buf
++ = ((char *)psz
)[0];
1354 len
+= sizeof(wxUint32
);
1358 if (buf
&& len
<=n
-sizeof(wxUint32
))
1368 // ============================================================================
1369 // The classes doing conversion using the iconv_xxx() functions
1370 // ============================================================================
1374 // VS: glibc 2.1.3 is broken in that iconv() conversion to/from UCS4 fails with
1375 // E2BIG if output buffer is _exactly_ as big as needed. Such case is
1376 // (unless there's yet another bug in glibc) the only case when iconv()
1377 // returns with (size_t)-1 (which means error) and says there are 0 bytes
1378 // left in the input buffer -- when _real_ error occurs,
1379 // bytes-left-in-input buffer is non-zero. Hence, this alternative test for
1381 // [This bug does not appear in glibc 2.2.]
1382 #if defined(__GLIBC__) && __GLIBC__ == 2 && __GLIBC_MINOR__ <= 1
1383 #define ICONV_FAILED(cres, bufLeft) ((cres == (size_t)-1) && \
1384 (errno != E2BIG || bufLeft != 0))
1386 #define ICONV_FAILED(cres, bufLeft) (cres == (size_t)-1)
1389 #define ICONV_CHAR_CAST(x) ((ICONV_CONST char **)(x))
1391 #define ICONV_T_INVALID ((iconv_t)-1)
1393 #if SIZEOF_WCHAR_T == 4
1394 #define WC_BSWAP wxUINT32_SWAP_ALWAYS
1395 #define WC_ENC wxFONTENCODING_UTF32
1396 #elif SIZEOF_WCHAR_T == 2
1397 #define WC_BSWAP wxUINT16_SWAP_ALWAYS
1398 #define WC_ENC wxFONTENCODING_UTF16
1399 #else // sizeof(wchar_t) != 2 nor 4
1400 // does this ever happen?
1401 #error "Unknown sizeof(wchar_t): please report this to wx-dev@lists.wxwindows.org"
1404 // ----------------------------------------------------------------------------
1405 // wxMBConv_iconv: encapsulates an iconv character set
1406 // ----------------------------------------------------------------------------
1408 class wxMBConv_iconv
: public wxMBConv
1411 wxMBConv_iconv(const wxChar
*name
);
1412 virtual ~wxMBConv_iconv();
1414 virtual size_t MB2WC(wchar_t *buf
, const char *psz
, size_t n
) const;
1415 virtual size_t WC2MB(char *buf
, const wchar_t *psz
, size_t n
) const;
1417 // classify this encoding as explained in wxMBConv::GetMBNulLen()
1419 virtual size_t GetMBNulLen() const;
1422 { return (m2w
!= ICONV_T_INVALID
) && (w2m
!= ICONV_T_INVALID
); }
1425 // the iconv handlers used to translate from multibyte to wide char and in
1426 // the other direction
1430 // guards access to m2w and w2m objects
1431 wxMutex m_iconvMutex
;
1435 // the name (for iconv_open()) of a wide char charset -- if none is
1436 // available on this machine, it will remain NULL
1437 static wxString ms_wcCharsetName
;
1439 // true if the wide char encoding we use (i.e. ms_wcCharsetName) has
1440 // different endian-ness than the native one
1441 static bool ms_wcNeedsSwap
;
1443 // cached result of GetMBNulLen(); set to 0 meaning "unknown"
1445 size_t m_minMBCharWidth
;
1448 // make the constructor available for unit testing
1449 WXDLLIMPEXP_BASE wxMBConv
* new_wxMBConv_iconv( const wxChar
* name
)
1451 wxMBConv_iconv
* result
= new wxMBConv_iconv( name
);
1452 if ( !result
->IsOk() )
1460 wxString
wxMBConv_iconv::ms_wcCharsetName
;
1461 bool wxMBConv_iconv::ms_wcNeedsSwap
= false;
1463 wxMBConv_iconv::wxMBConv_iconv(const wxChar
*name
)
1465 m_minMBCharWidth
= 0;
1467 // iconv operates with chars, not wxChars, but luckily it uses only ASCII
1468 // names for the charsets
1469 const wxCharBuffer
cname(wxString(name
).ToAscii());
1471 // check for charset that represents wchar_t:
1472 if ( ms_wcCharsetName
.empty() )
1474 wxLogTrace(TRACE_STRCONV
, _T("Looking for wide char codeset:"));
1477 const wxChar
**names
= wxFontMapperBase::GetAllEncodingNames(WC_ENC
);
1478 #else // !wxUSE_FONTMAP
1479 static const wxChar
*names
[] =
1481 #if SIZEOF_WCHAR_T == 4
1483 #elif SIZEOF_WCHAR_T = 2
1488 #endif // wxUSE_FONTMAP/!wxUSE_FONTMAP
1490 for ( ; *names
&& ms_wcCharsetName
.empty(); ++names
)
1492 const wxString
nameCS(*names
);
1494 // first try charset with explicit bytesex info (e.g. "UCS-4LE"):
1495 wxString
nameXE(nameCS
);
1496 #ifdef WORDS_BIGENDIAN
1498 #else // little endian
1502 wxLogTrace(TRACE_STRCONV
, _T(" trying charset \"%s\""),
1505 m2w
= iconv_open(nameXE
.ToAscii(), cname
);
1506 if ( m2w
== ICONV_T_INVALID
)
1508 // try charset w/o bytesex info (e.g. "UCS4")
1509 wxLogTrace(TRACE_STRCONV
, _T(" trying charset \"%s\""),
1511 m2w
= iconv_open(nameCS
.ToAscii(), cname
);
1513 // and check for bytesex ourselves:
1514 if ( m2w
!= ICONV_T_INVALID
)
1516 char buf
[2], *bufPtr
;
1517 wchar_t wbuf
[2], *wbufPtr
;
1525 outsz
= SIZEOF_WCHAR_T
* 2;
1529 res
= iconv(m2w
, ICONV_CHAR_CAST(&bufPtr
), &insz
,
1530 (char**)&wbufPtr
, &outsz
);
1532 if (ICONV_FAILED(res
, insz
))
1534 wxLogLastError(wxT("iconv"));
1535 wxLogError(_("Conversion to charset '%s' doesn't work."),
1538 else // ok, can convert to this encoding, remember it
1540 ms_wcCharsetName
= nameCS
;
1541 ms_wcNeedsSwap
= wbuf
[0] != (wchar_t)buf
[0];
1545 else // use charset not requiring byte swapping
1547 ms_wcCharsetName
= nameXE
;
1551 wxLogTrace(TRACE_STRCONV
,
1552 wxT("iconv wchar_t charset is \"%s\"%s"),
1553 ms_wcCharsetName
.empty() ? _T("<none>")
1554 : ms_wcCharsetName
.c_str(),
1555 ms_wcNeedsSwap
? _T(" (needs swap)")
1558 else // we already have ms_wcCharsetName
1560 m2w
= iconv_open(ms_wcCharsetName
.ToAscii(), cname
);
1563 if ( ms_wcCharsetName
.empty() )
1565 w2m
= ICONV_T_INVALID
;
1569 w2m
= iconv_open(cname
, ms_wcCharsetName
.ToAscii());
1570 if ( w2m
== ICONV_T_INVALID
)
1572 wxLogTrace(TRACE_STRCONV
,
1573 wxT("\"%s\" -> \"%s\" works but not the converse!?"),
1574 ms_wcCharsetName
.c_str(), cname
.data());
1579 wxMBConv_iconv::~wxMBConv_iconv()
1581 if ( m2w
!= ICONV_T_INVALID
)
1583 if ( w2m
!= ICONV_T_INVALID
)
1587 size_t wxMBConv_iconv::MB2WC(wchar_t *buf
, const char *psz
, size_t n
) const
1589 // find the string length: notice that must be done differently for
1590 // NUL-terminated strings and UTF-16/32 which are terminated with 2/4 NULs
1592 const size_t nulLen
= GetMBNulLen();
1599 inbuf
= strlen(psz
); // arguably more optimized than our version
1604 // for UTF-16/32 not only we need to have 2/4 consecutive NULs but
1605 // they also have to start at character boundary and not span two
1606 // adjacent characters
1608 for ( p
= psz
; NotAllNULs(p
, nulLen
); p
+= nulLen
)
1615 // NB: iconv() is MT-safe, but each thread must use it's own iconv_t handle.
1616 // Unfortunately there is a couple of global wxCSConv objects such as
1617 // wxConvLocal that are used all over wx code, so we have to make sure
1618 // the handle is used by at most one thread at the time. Otherwise
1619 // only a few wx classes would be safe to use from non-main threads
1620 // as MB<->WC conversion would fail "randomly".
1621 wxMutexLocker
lock(wxConstCast(this, wxMBConv_iconv
)->m_iconvMutex
);
1622 #endif // wxUSE_THREADS
1625 size_t outbuf
= n
* SIZEOF_WCHAR_T
;
1627 // VS: Use these instead of psz, buf because iconv() modifies its arguments:
1628 wchar_t *bufPtr
= buf
;
1629 const char *pszPtr
= psz
;
1633 // have destination buffer, convert there
1635 ICONV_CHAR_CAST(&pszPtr
), &inbuf
,
1636 (char**)&bufPtr
, &outbuf
);
1637 res
= n
- (outbuf
/ SIZEOF_WCHAR_T
);
1641 // convert to native endianness
1642 for ( unsigned i
= 0; i
< res
; i
++ )
1643 buf
[n
] = WC_BSWAP(buf
[i
]);
1646 // NUL-terminate the string if there is any space left
1652 // no destination buffer... convert using temp buffer
1653 // to calculate destination buffer requirement
1658 outbuf
= 8*SIZEOF_WCHAR_T
;
1661 ICONV_CHAR_CAST(&pszPtr
), &inbuf
,
1662 (char**)&bufPtr
, &outbuf
);
1664 res
+= 8-(outbuf
/SIZEOF_WCHAR_T
);
1665 } while ((cres
==(size_t)-1) && (errno
==E2BIG
));
1668 if (ICONV_FAILED(cres
, inbuf
))
1670 //VS: it is ok if iconv fails, hence trace only
1671 wxLogTrace(TRACE_STRCONV
, wxT("iconv failed: %s"), wxSysErrorMsg(wxSysErrorCode()));
1678 size_t wxMBConv_iconv::WC2MB(char *buf
, const wchar_t *psz
, size_t n
) const
1681 // NB: explained in MB2WC
1682 wxMutexLocker
lock(wxConstCast(this, wxMBConv_iconv
)->m_iconvMutex
);
1685 size_t inlen
= wxWcslen(psz
);
1686 size_t inbuf
= inlen
* SIZEOF_WCHAR_T
;
1690 wchar_t *tmpbuf
= 0;
1694 // need to copy to temp buffer to switch endianness
1695 // (doing WC_BSWAP twice on the original buffer won't help, as it
1696 // could be in read-only memory, or be accessed in some other thread)
1697 tmpbuf
= (wchar_t *)malloc(inbuf
+ SIZEOF_WCHAR_T
);
1698 for ( size_t i
= 0; i
< inlen
; i
++ )
1699 tmpbuf
[n
] = WC_BSWAP(psz
[i
]);
1700 tmpbuf
[inlen
] = L
'\0';
1706 // have destination buffer, convert there
1707 cres
= iconv( w2m
, ICONV_CHAR_CAST(&psz
), &inbuf
, &buf
, &outbuf
);
1711 // NB: iconv was given only wcslen(psz) characters on input, and so
1712 // it couldn't convert the trailing zero. Let's do it ourselves
1713 // if there's some room left for it in the output buffer.
1719 // no destination buffer... convert using temp buffer
1720 // to calculate destination buffer requirement
1724 buf
= tbuf
; outbuf
= 16;
1726 cres
= iconv( w2m
, ICONV_CHAR_CAST(&psz
), &inbuf
, &buf
, &outbuf
);
1729 } while ((cres
==(size_t)-1) && (errno
==E2BIG
));
1737 if (ICONV_FAILED(cres
, inbuf
))
1739 wxLogTrace(TRACE_STRCONV
, wxT("iconv failed: %s"), wxSysErrorMsg(wxSysErrorCode()));
1746 size_t wxMBConv_iconv::GetMBNulLen() const
1748 if ( m_minMBCharWidth
== 0 )
1750 wxMBConv_iconv
* const self
= wxConstCast(this, wxMBConv_iconv
);
1753 // NB: explained in MB2WC
1754 wxMutexLocker
lock(self
->m_iconvMutex
);
1757 wchar_t *wnul
= L
"";
1758 char buf
[8]; // should be enough for NUL in any encoding
1759 size_t inLen
= sizeof(wchar_t),
1760 outLen
= WXSIZEOF(buf
);
1761 char *in
= (char *)wnul
;
1763 if ( iconv(w2m
, ICONV_CHAR_CAST(&in
), &inLen
, &out
, &outLen
) == (size_t)-1 )
1765 self
->m_minMBCharWidth
= (size_t)-1;
1769 self
->m_minMBCharWidth
= out
- buf
;
1773 return m_minMBCharWidth
;
1776 #endif // HAVE_ICONV
1779 // ============================================================================
1780 // Win32 conversion classes
1781 // ============================================================================
1783 #ifdef wxHAVE_WIN32_MB2WC
1787 extern WXDLLIMPEXP_BASE
long wxCharsetToCodepage(const wxChar
*charset
);
1788 extern WXDLLIMPEXP_BASE
long wxEncodingToCodepage(wxFontEncoding encoding
);
1791 class wxMBConv_win32
: public wxMBConv
1796 m_CodePage
= CP_ACP
;
1797 m_minMBCharWidth
= 0;
1801 wxMBConv_win32(const wxChar
* name
)
1803 m_CodePage
= wxCharsetToCodepage(name
);
1804 m_minMBCharWidth
= 0;
1807 wxMBConv_win32(wxFontEncoding encoding
)
1809 m_CodePage
= wxEncodingToCodepage(encoding
);
1810 m_minMBCharWidth
= 0;
1812 #endif // wxUSE_FONTMAP
1814 size_t MB2WC(wchar_t *buf
, const char *psz
, size_t n
) const
1816 // note that we have to use MB_ERR_INVALID_CHARS flag as it without it
1817 // the behaviour is not compatible with the Unix version (using iconv)
1818 // and break the library itself, e.g. wxTextInputStream::NextChar()
1819 // wouldn't work if reading an incomplete MB char didn't result in an
1822 // Moreover, MB_ERR_INVALID_CHARS is only supported on Win 2K SP4 or
1823 // Win XP or newer and it is not supported for UTF-[78] so we always
1824 // use our own conversions in this case. See
1825 // http://blogs.msdn.com/michkap/archive/2005/04/19/409566.aspx
1826 // http://msdn.microsoft.com/library/en-us/intl/unicode_17si.asp
1827 if ( m_CodePage
== CP_UTF8
)
1829 return wxConvUTF8
.MB2WC(buf
, psz
, n
);
1832 if ( m_CodePage
== CP_UTF7
)
1834 return wxConvUTF7
.MB2WC(buf
, psz
, n
);
1838 if ( (m_CodePage
< 50000 && m_CodePage
!= CP_SYMBOL
) &&
1839 IsAtLeastWin2kSP4() )
1841 flags
= MB_ERR_INVALID_CHARS
;
1844 const size_t len
= ::MultiByteToWideChar
1846 m_CodePage
, // code page
1847 flags
, // flags: fall on error
1848 psz
, // input string
1849 -1, // its length (NUL-terminated)
1850 buf
, // output string
1851 buf
? n
: 0 // size of output buffer
1855 // function totally failed
1859 // if we were really converting and didn't use MB_ERR_INVALID_CHARS,
1860 // check if we succeeded, by doing a double trip:
1861 if ( !flags
&& buf
)
1863 const size_t mbLen
= strlen(psz
);
1864 wxCharBuffer
mbBuf(mbLen
);
1865 if ( ::WideCharToMultiByte
1872 mbLen
+ 1, // size in bytes, not length
1876 strcmp(mbBuf
, psz
) != 0 )
1878 // we didn't obtain the same thing we started from, hence
1879 // the conversion was lossy and we consider that it failed
1884 // note that it returns count of written chars for buf != NULL and size
1885 // of the needed buffer for buf == NULL so in either case the length of
1886 // the string (which never includes the terminating NUL) is one less
1890 size_t WC2MB(char *buf
, const wchar_t *pwz
, size_t n
) const
1893 we have a problem here: by default, WideCharToMultiByte() may
1894 replace characters unrepresentable in the target code page with bad
1895 quality approximations such as turning "1/2" symbol (U+00BD) into
1896 "1" for the code pages which don't have it and we, obviously, want
1897 to avoid this at any price
1899 the trouble is that this function does it _silently_, i.e. it won't
1900 even tell us whether it did or not... Win98/2000 and higher provide
1901 WC_NO_BEST_FIT_CHARS but it doesn't work for the older systems and
1902 we have to resort to a round trip, i.e. check that converting back
1903 results in the same string -- this is, of course, expensive but
1904 otherwise we simply can't be sure to not garble the data.
1907 // determine if we can rely on WC_NO_BEST_FIT_CHARS: according to MSDN
1908 // it doesn't work with CJK encodings (which we test for rather roughly
1909 // here...) nor with UTF-7/8 nor, of course, with Windows versions not
1911 BOOL usedDef
wxDUMMY_INITIALIZE(false);
1914 if ( CanUseNoBestFit() && m_CodePage
< 50000 )
1916 // it's our lucky day
1917 flags
= WC_NO_BEST_FIT_CHARS
;
1918 pUsedDef
= &usedDef
;
1920 else // old system or unsupported encoding
1926 const size_t len
= ::WideCharToMultiByte
1928 m_CodePage
, // code page
1929 flags
, // either none or no best fit
1930 pwz
, // input string
1931 -1, // it is (wide) NUL-terminated
1932 buf
, // output buffer
1933 buf
? n
: 0, // and its size
1934 NULL
, // default "replacement" char
1935 pUsedDef
// [out] was it used?
1940 // function totally failed
1944 // if we were really converting, check if we succeeded
1949 // check if the conversion failed, i.e. if any replacements
1954 else // we must resort to double tripping...
1956 wxWCharBuffer
wcBuf(n
);
1957 if ( MB2WC(wcBuf
.data(), buf
, n
) == (size_t)-1 ||
1958 wcscmp(wcBuf
, pwz
) != 0 )
1960 // we didn't obtain the same thing we started from, hence
1961 // the conversion was lossy and we consider that it failed
1967 // see the comment above for the reason of "len - 1"
1971 virtual size_t GetMBNulLen() const
1973 if ( m_minMBCharWidth
== 0 )
1975 int len
= ::WideCharToMultiByte
1977 m_CodePage
, // code page
1979 L
"", // input string
1980 1, // translate just the NUL
1981 NULL
, // output buffer
1983 NULL
, // no replacement char
1984 NULL
// [out] don't care if it was used
1987 wxMBConv_win32
* const self
= wxConstCast(this, wxMBConv_win32
);
1991 wxLogDebug(_T("Unexpected NUL length %d"), len
);
1995 self
->m_minMBCharWidth
= (size_t)-1;
2001 self
->m_minMBCharWidth
= len
;
2006 return m_minMBCharWidth
;
2009 bool IsOk() const { return m_CodePage
!= -1; }
2012 static bool CanUseNoBestFit()
2014 static int s_isWin98Or2k
= -1;
2016 if ( s_isWin98Or2k
== -1 )
2019 switch ( wxGetOsVersion(&verMaj
, &verMin
) )
2022 s_isWin98Or2k
= verMaj
>= 4 && verMin
>= 10;
2026 s_isWin98Or2k
= verMaj
>= 5;
2030 // unknown, be conseravtive by default
2034 wxASSERT_MSG( s_isWin98Or2k
!= -1, _T("should be set above") );
2037 return s_isWin98Or2k
== 1;
2040 static bool IsAtLeastWin2kSP4()
2045 static int s_isAtLeastWin2kSP4
= -1;
2047 if ( s_isAtLeastWin2kSP4
== -1 )
2049 OSVERSIONINFOEX ver
;
2051 memset(&ver
, 0, sizeof(ver
));
2052 ver
.dwOSVersionInfoSize
= sizeof(ver
);
2053 GetVersionEx((OSVERSIONINFO
*)&ver
);
2055 s_isAtLeastWin2kSP4
=
2056 ((ver
.dwMajorVersion
> 5) || // Vista+
2057 (ver
.dwMajorVersion
== 5 && ver
.dwMinorVersion
> 0) || // XP/2003
2058 (ver
.dwMajorVersion
== 5 && ver
.dwMinorVersion
== 0 &&
2059 ver
.wServicePackMajor
>= 4)) // 2000 SP4+
2063 return s_isAtLeastWin2kSP4
== 1;
2068 // the code page we're working with
2071 // cached result of GetMBNulLen(), set to 0 initially meaning
2073 size_t m_minMBCharWidth
;
2076 #endif // wxHAVE_WIN32_MB2WC
2078 // ============================================================================
2079 // Cocoa conversion classes
2080 // ============================================================================
2082 #if defined(__WXCOCOA__)
2084 // RN: There is no UTF-32 support in either Core Foundation or
2085 // Cocoa. Strangely enough, internally Core Foundation uses
2086 // UTF 32 internally quite a bit - its just not public (yet).
2088 #include <CoreFoundation/CFString.h>
2089 #include <CoreFoundation/CFStringEncodingExt.h>
2091 CFStringEncoding
wxCFStringEncFromFontEnc(wxFontEncoding encoding
)
2093 CFStringEncoding enc
= kCFStringEncodingInvalidId
;
2094 if ( encoding
== wxFONTENCODING_DEFAULT
)
2096 enc
= CFStringGetSystemEncoding();
2098 else switch( encoding
)
2100 case wxFONTENCODING_ISO8859_1
:
2101 enc
= kCFStringEncodingISOLatin1
;
2103 case wxFONTENCODING_ISO8859_2
:
2104 enc
= kCFStringEncodingISOLatin2
;
2106 case wxFONTENCODING_ISO8859_3
:
2107 enc
= kCFStringEncodingISOLatin3
;
2109 case wxFONTENCODING_ISO8859_4
:
2110 enc
= kCFStringEncodingISOLatin4
;
2112 case wxFONTENCODING_ISO8859_5
:
2113 enc
= kCFStringEncodingISOLatinCyrillic
;
2115 case wxFONTENCODING_ISO8859_6
:
2116 enc
= kCFStringEncodingISOLatinArabic
;
2118 case wxFONTENCODING_ISO8859_7
:
2119 enc
= kCFStringEncodingISOLatinGreek
;
2121 case wxFONTENCODING_ISO8859_8
:
2122 enc
= kCFStringEncodingISOLatinHebrew
;
2124 case wxFONTENCODING_ISO8859_9
:
2125 enc
= kCFStringEncodingISOLatin5
;
2127 case wxFONTENCODING_ISO8859_10
:
2128 enc
= kCFStringEncodingISOLatin6
;
2130 case wxFONTENCODING_ISO8859_11
:
2131 enc
= kCFStringEncodingISOLatinThai
;
2133 case wxFONTENCODING_ISO8859_13
:
2134 enc
= kCFStringEncodingISOLatin7
;
2136 case wxFONTENCODING_ISO8859_14
:
2137 enc
= kCFStringEncodingISOLatin8
;
2139 case wxFONTENCODING_ISO8859_15
:
2140 enc
= kCFStringEncodingISOLatin9
;
2143 case wxFONTENCODING_KOI8
:
2144 enc
= kCFStringEncodingKOI8_R
;
2146 case wxFONTENCODING_ALTERNATIVE
: // MS-DOS CP866
2147 enc
= kCFStringEncodingDOSRussian
;
2150 // case wxFONTENCODING_BULGARIAN :
2154 case wxFONTENCODING_CP437
:
2155 enc
=kCFStringEncodingDOSLatinUS
;
2157 case wxFONTENCODING_CP850
:
2158 enc
= kCFStringEncodingDOSLatin1
;
2160 case wxFONTENCODING_CP852
:
2161 enc
= kCFStringEncodingDOSLatin2
;
2163 case wxFONTENCODING_CP855
:
2164 enc
= kCFStringEncodingDOSCyrillic
;
2166 case wxFONTENCODING_CP866
:
2167 enc
=kCFStringEncodingDOSRussian
;
2169 case wxFONTENCODING_CP874
:
2170 enc
= kCFStringEncodingDOSThai
;
2172 case wxFONTENCODING_CP932
:
2173 enc
= kCFStringEncodingDOSJapanese
;
2175 case wxFONTENCODING_CP936
:
2176 enc
=kCFStringEncodingDOSChineseSimplif
;
2178 case wxFONTENCODING_CP949
:
2179 enc
= kCFStringEncodingDOSKorean
;
2181 case wxFONTENCODING_CP950
:
2182 enc
= kCFStringEncodingDOSChineseTrad
;
2184 case wxFONTENCODING_CP1250
:
2185 enc
= kCFStringEncodingWindowsLatin2
;
2187 case wxFONTENCODING_CP1251
:
2188 enc
=kCFStringEncodingWindowsCyrillic
;
2190 case wxFONTENCODING_CP1252
:
2191 enc
=kCFStringEncodingWindowsLatin1
;
2193 case wxFONTENCODING_CP1253
:
2194 enc
= kCFStringEncodingWindowsGreek
;
2196 case wxFONTENCODING_CP1254
:
2197 enc
= kCFStringEncodingWindowsLatin5
;
2199 case wxFONTENCODING_CP1255
:
2200 enc
=kCFStringEncodingWindowsHebrew
;
2202 case wxFONTENCODING_CP1256
:
2203 enc
=kCFStringEncodingWindowsArabic
;
2205 case wxFONTENCODING_CP1257
:
2206 enc
= kCFStringEncodingWindowsBalticRim
;
2208 // This only really encodes to UTF7 (if that) evidently
2209 // case wxFONTENCODING_UTF7 :
2210 // enc = kCFStringEncodingNonLossyASCII ;
2212 case wxFONTENCODING_UTF8
:
2213 enc
= kCFStringEncodingUTF8
;
2215 case wxFONTENCODING_EUC_JP
:
2216 enc
= kCFStringEncodingEUC_JP
;
2218 case wxFONTENCODING_UTF16
:
2219 enc
= kCFStringEncodingUnicode
;
2221 case wxFONTENCODING_MACROMAN
:
2222 enc
= kCFStringEncodingMacRoman
;
2224 case wxFONTENCODING_MACJAPANESE
:
2225 enc
= kCFStringEncodingMacJapanese
;
2227 case wxFONTENCODING_MACCHINESETRAD
:
2228 enc
= kCFStringEncodingMacChineseTrad
;
2230 case wxFONTENCODING_MACKOREAN
:
2231 enc
= kCFStringEncodingMacKorean
;
2233 case wxFONTENCODING_MACARABIC
:
2234 enc
= kCFStringEncodingMacArabic
;
2236 case wxFONTENCODING_MACHEBREW
:
2237 enc
= kCFStringEncodingMacHebrew
;
2239 case wxFONTENCODING_MACGREEK
:
2240 enc
= kCFStringEncodingMacGreek
;
2242 case wxFONTENCODING_MACCYRILLIC
:
2243 enc
= kCFStringEncodingMacCyrillic
;
2245 case wxFONTENCODING_MACDEVANAGARI
:
2246 enc
= kCFStringEncodingMacDevanagari
;
2248 case wxFONTENCODING_MACGURMUKHI
:
2249 enc
= kCFStringEncodingMacGurmukhi
;
2251 case wxFONTENCODING_MACGUJARATI
:
2252 enc
= kCFStringEncodingMacGujarati
;
2254 case wxFONTENCODING_MACORIYA
:
2255 enc
= kCFStringEncodingMacOriya
;
2257 case wxFONTENCODING_MACBENGALI
:
2258 enc
= kCFStringEncodingMacBengali
;
2260 case wxFONTENCODING_MACTAMIL
:
2261 enc
= kCFStringEncodingMacTamil
;
2263 case wxFONTENCODING_MACTELUGU
:
2264 enc
= kCFStringEncodingMacTelugu
;
2266 case wxFONTENCODING_MACKANNADA
:
2267 enc
= kCFStringEncodingMacKannada
;
2269 case wxFONTENCODING_MACMALAJALAM
:
2270 enc
= kCFStringEncodingMacMalayalam
;
2272 case wxFONTENCODING_MACSINHALESE
:
2273 enc
= kCFStringEncodingMacSinhalese
;
2275 case wxFONTENCODING_MACBURMESE
:
2276 enc
= kCFStringEncodingMacBurmese
;
2278 case wxFONTENCODING_MACKHMER
:
2279 enc
= kCFStringEncodingMacKhmer
;
2281 case wxFONTENCODING_MACTHAI
:
2282 enc
= kCFStringEncodingMacThai
;
2284 case wxFONTENCODING_MACLAOTIAN
:
2285 enc
= kCFStringEncodingMacLaotian
;
2287 case wxFONTENCODING_MACGEORGIAN
:
2288 enc
= kCFStringEncodingMacGeorgian
;
2290 case wxFONTENCODING_MACARMENIAN
:
2291 enc
= kCFStringEncodingMacArmenian
;
2293 case wxFONTENCODING_MACCHINESESIMP
:
2294 enc
= kCFStringEncodingMacChineseSimp
;
2296 case wxFONTENCODING_MACTIBETAN
:
2297 enc
= kCFStringEncodingMacTibetan
;
2299 case wxFONTENCODING_MACMONGOLIAN
:
2300 enc
= kCFStringEncodingMacMongolian
;
2302 case wxFONTENCODING_MACETHIOPIC
:
2303 enc
= kCFStringEncodingMacEthiopic
;
2305 case wxFONTENCODING_MACCENTRALEUR
:
2306 enc
= kCFStringEncodingMacCentralEurRoman
;
2308 case wxFONTENCODING_MACVIATNAMESE
:
2309 enc
= kCFStringEncodingMacVietnamese
;
2311 case wxFONTENCODING_MACARABICEXT
:
2312 enc
= kCFStringEncodingMacExtArabic
;
2314 case wxFONTENCODING_MACSYMBOL
:
2315 enc
= kCFStringEncodingMacSymbol
;
2317 case wxFONTENCODING_MACDINGBATS
:
2318 enc
= kCFStringEncodingMacDingbats
;
2320 case wxFONTENCODING_MACTURKISH
:
2321 enc
= kCFStringEncodingMacTurkish
;
2323 case wxFONTENCODING_MACCROATIAN
:
2324 enc
= kCFStringEncodingMacCroatian
;
2326 case wxFONTENCODING_MACICELANDIC
:
2327 enc
= kCFStringEncodingMacIcelandic
;
2329 case wxFONTENCODING_MACROMANIAN
:
2330 enc
= kCFStringEncodingMacRomanian
;
2332 case wxFONTENCODING_MACCELTIC
:
2333 enc
= kCFStringEncodingMacCeltic
;
2335 case wxFONTENCODING_MACGAELIC
:
2336 enc
= kCFStringEncodingMacGaelic
;
2338 // case wxFONTENCODING_MACKEYBOARD :
2339 // enc = kCFStringEncodingMacKeyboardGlyphs ;
2342 // because gcc is picky
2348 class wxMBConv_cocoa
: public wxMBConv
2353 Init(CFStringGetSystemEncoding()) ;
2357 wxMBConv_cocoa(const wxChar
* name
)
2359 Init( wxCFStringEncFromFontEnc(wxFontMapperBase::Get()->CharsetToEncoding(name
, false) ) ) ;
2363 wxMBConv_cocoa(wxFontEncoding encoding
)
2365 Init( wxCFStringEncFromFontEnc(encoding
) );
2372 void Init( CFStringEncoding encoding
)
2374 m_encoding
= encoding
;
2377 size_t MB2WC(wchar_t * szOut
, const char * szUnConv
, size_t nOutSize
) const
2381 CFStringRef theString
= CFStringCreateWithBytes (
2382 NULL
, //the allocator
2383 (const UInt8
*)szUnConv
,
2386 false //no BOM/external representation
2389 wxASSERT(theString
);
2391 size_t nOutLength
= CFStringGetLength(theString
);
2395 CFRelease(theString
);
2399 CFRange theRange
= { 0, nOutSize
};
2401 #if SIZEOF_WCHAR_T == 4
2402 UniChar
* szUniCharBuffer
= new UniChar
[nOutSize
];
2405 CFStringGetCharacters(theString
, theRange
, szUniCharBuffer
);
2407 CFRelease(theString
);
2409 szUniCharBuffer
[nOutLength
] = '\0' ;
2411 #if SIZEOF_WCHAR_T == 4
2412 wxMBConvUTF16 converter
;
2413 converter
.MB2WC(szOut
, (const char*)szUniCharBuffer
, nOutSize
) ;
2414 delete[] szUniCharBuffer
;
2420 size_t WC2MB(char *szOut
, const wchar_t *szUnConv
, size_t nOutSize
) const
2424 size_t nRealOutSize
;
2425 size_t nBufSize
= wxWcslen(szUnConv
);
2426 UniChar
* szUniBuffer
= (UniChar
*) szUnConv
;
2428 #if SIZEOF_WCHAR_T == 4
2429 wxMBConvUTF16 converter
;
2430 nBufSize
= converter
.WC2MB( NULL
, szUnConv
, 0 );
2431 szUniBuffer
= new UniChar
[ (nBufSize
/ sizeof(UniChar
)) + 1] ;
2432 converter
.WC2MB( (char*) szUniBuffer
, szUnConv
, nBufSize
+ sizeof(UniChar
)) ;
2433 nBufSize
/= sizeof(UniChar
);
2436 CFStringRef theString
= CFStringCreateWithCharactersNoCopy(
2440 kCFAllocatorNull
//deallocator - we want to deallocate it ourselves
2443 wxASSERT(theString
);
2445 //Note that CER puts a BOM when converting to unicode
2446 //so we check and use getchars instead in that case
2447 if (m_encoding
== kCFStringEncodingUnicode
)
2450 CFStringGetCharacters(theString
, CFRangeMake(0, nOutSize
- 1), (UniChar
*) szOut
);
2452 nRealOutSize
= CFStringGetLength(theString
) + 1;
2458 CFRangeMake(0, CFStringGetLength(theString
)),
2460 0, //what to put in characters that can't be converted -
2461 //0 tells CFString to return NULL if it meets such a character
2462 false, //not an external representation
2465 (CFIndex
*) &nRealOutSize
2469 CFRelease(theString
);
2471 #if SIZEOF_WCHAR_T == 4
2472 delete[] szUniBuffer
;
2475 return nRealOutSize
- 1;
2480 return m_encoding
!= kCFStringEncodingInvalidId
&&
2481 CFStringIsEncodingAvailable(m_encoding
);
2485 CFStringEncoding m_encoding
;
2488 #endif // defined(__WXCOCOA__)
2490 // ============================================================================
2491 // Mac conversion classes
2492 // ============================================================================
2494 #if defined(__WXMAC__) && defined(TARGET_CARBON)
2496 class wxMBConv_mac
: public wxMBConv
2501 Init(CFStringGetSystemEncoding()) ;
2505 wxMBConv_mac(const wxChar
* name
)
2507 Init( wxMacGetSystemEncFromFontEnc(wxFontMapperBase::Get()->CharsetToEncoding(name
, false) ) ) ;
2511 wxMBConv_mac(wxFontEncoding encoding
)
2513 Init( wxMacGetSystemEncFromFontEnc(encoding
) );
2518 OSStatus status
= noErr
;
2519 status
= TECDisposeConverter(m_MB2WC_converter
);
2520 status
= TECDisposeConverter(m_WC2MB_converter
);
2524 void Init( TextEncodingBase encoding
)
2526 OSStatus status
= noErr
;
2527 m_char_encoding
= encoding
;
2528 m_unicode_encoding
= CreateTextEncoding(kTextEncodingUnicodeDefault
,0,kUnicode16BitFormat
) ;
2530 status
= TECCreateConverter(&m_MB2WC_converter
,
2532 m_unicode_encoding
);
2533 status
= TECCreateConverter(&m_WC2MB_converter
,
2538 size_t MB2WC(wchar_t *buf
, const char *psz
, size_t n
) const
2540 OSStatus status
= noErr
;
2541 ByteCount byteOutLen
;
2542 ByteCount byteInLen
= strlen(psz
) ;
2543 wchar_t *tbuf
= NULL
;
2544 UniChar
* ubuf
= NULL
;
2549 //apple specs say at least 32
2550 n
= wxMax( 32 , byteInLen
) ;
2551 tbuf
= (wchar_t*) malloc( n
* SIZEOF_WCHAR_T
) ;
2553 ByteCount byteBufferLen
= n
* sizeof( UniChar
) ;
2554 #if SIZEOF_WCHAR_T == 4
2555 ubuf
= (UniChar
*) malloc( byteBufferLen
+ 2 ) ;
2557 ubuf
= (UniChar
*) (buf
? buf
: tbuf
) ;
2559 status
= TECConvertText(m_MB2WC_converter
, (ConstTextPtr
) psz
, byteInLen
, &byteInLen
,
2560 (TextPtr
) ubuf
, byteBufferLen
, &byteOutLen
);
2561 #if SIZEOF_WCHAR_T == 4
2562 // we have to terminate here, because n might be larger for the trailing zero, and if UniChar
2563 // is not properly terminated we get random characters at the end
2564 ubuf
[byteOutLen
/ sizeof( UniChar
) ] = 0 ;
2565 wxMBConvUTF16 converter
;
2566 res
= converter
.MB2WC( (buf
? buf
: tbuf
) , (const char*)ubuf
, n
) ;
2569 res
= byteOutLen
/ sizeof( UniChar
) ;
2574 if ( buf
&& res
< n
)
2580 size_t WC2MB(char *buf
, const wchar_t *psz
, size_t n
) const
2582 OSStatus status
= noErr
;
2583 ByteCount byteOutLen
;
2584 ByteCount byteInLen
= wxWcslen(psz
) * SIZEOF_WCHAR_T
;
2590 //apple specs say at least 32
2591 n
= wxMax( 32 , ((byteInLen
/ SIZEOF_WCHAR_T
) * 8) + SIZEOF_WCHAR_T
);
2592 tbuf
= (char*) malloc( n
) ;
2595 ByteCount byteBufferLen
= n
;
2596 UniChar
* ubuf
= NULL
;
2597 #if SIZEOF_WCHAR_T == 4
2598 wxMBConvUTF16 converter
;
2599 size_t unicharlen
= converter
.WC2MB( NULL
, psz
, 0 ) ;
2600 byteInLen
= unicharlen
;
2601 ubuf
= (UniChar
*) malloc( byteInLen
+ 2 ) ;
2602 converter
.WC2MB( (char*) ubuf
, psz
, unicharlen
+ 2 ) ;
2604 ubuf
= (UniChar
*) psz
;
2606 status
= TECConvertText(m_WC2MB_converter
, (ConstTextPtr
) ubuf
, byteInLen
, &byteInLen
,
2607 (TextPtr
) (buf
? buf
: tbuf
) , byteBufferLen
, &byteOutLen
);
2608 #if SIZEOF_WCHAR_T == 4
2614 size_t res
= byteOutLen
;
2615 if ( buf
&& res
< n
)
2619 //we need to double-trip to verify it didn't insert any ? in place
2620 //of bogus characters
2621 wxWCharBuffer
wcBuf(n
);
2622 size_t pszlen
= wxWcslen(psz
);
2623 if ( MB2WC(wcBuf
.data(), buf
, n
) == (size_t)-1 ||
2624 wxWcslen(wcBuf
) != pszlen
||
2625 memcmp(wcBuf
, psz
, pszlen
* sizeof(wchar_t)) != 0 )
2627 // we didn't obtain the same thing we started from, hence
2628 // the conversion was lossy and we consider that it failed
2637 { return m_MB2WC_converter
!= NULL
&& m_WC2MB_converter
!= NULL
; }
2640 TECObjectRef m_MB2WC_converter
;
2641 TECObjectRef m_WC2MB_converter
;
2643 TextEncodingBase m_char_encoding
;
2644 TextEncodingBase m_unicode_encoding
;
2647 #endif // defined(__WXMAC__) && defined(TARGET_CARBON)
2649 // ============================================================================
2650 // wxEncodingConverter based conversion classes
2651 // ============================================================================
2655 class wxMBConv_wxwin
: public wxMBConv
2660 m_ok
= m2w
.Init(m_enc
, wxFONTENCODING_UNICODE
) &&
2661 w2m
.Init(wxFONTENCODING_UNICODE
, m_enc
);
2665 // temporarily just use wxEncodingConverter stuff,
2666 // so that it works while a better implementation is built
2667 wxMBConv_wxwin(const wxChar
* name
)
2670 m_enc
= wxFontMapperBase::Get()->CharsetToEncoding(name
, false);
2672 m_enc
= wxFONTENCODING_SYSTEM
;
2677 wxMBConv_wxwin(wxFontEncoding enc
)
2684 size_t MB2WC(wchar_t *buf
, const char *psz
, size_t WXUNUSED(n
)) const
2686 size_t inbuf
= strlen(psz
);
2689 if (!m2w
.Convert(psz
,buf
))
2695 size_t WC2MB(char *buf
, const wchar_t *psz
, size_t WXUNUSED(n
)) const
2697 const size_t inbuf
= wxWcslen(psz
);
2700 if (!w2m
.Convert(psz
,buf
))
2707 virtual size_t GetMBNulLen() const
2711 case wxFONTENCODING_UTF16BE
:
2712 case wxFONTENCODING_UTF16LE
:
2715 case wxFONTENCODING_UTF32BE
:
2716 case wxFONTENCODING_UTF32LE
:
2724 bool IsOk() const { return m_ok
; }
2727 wxFontEncoding m_enc
;
2728 wxEncodingConverter m2w
, w2m
;
2731 // were we initialized successfully?
2734 DECLARE_NO_COPY_CLASS(wxMBConv_wxwin
)
2737 // make the constructors available for unit testing
2738 WXDLLIMPEXP_BASE wxMBConv
* new_wxMBConv_wxwin( const wxChar
* name
)
2740 wxMBConv_wxwin
* result
= new wxMBConv_wxwin( name
);
2741 if ( !result
->IsOk() )
2749 #endif // wxUSE_FONTMAP
2751 // ============================================================================
2752 // wxCSConv implementation
2753 // ============================================================================
2755 void wxCSConv::Init()
2762 wxCSConv::wxCSConv(const wxChar
*charset
)
2772 m_encoding
= wxFontMapperBase::GetEncodingFromName(charset
);
2774 m_encoding
= wxFONTENCODING_SYSTEM
;
2778 wxCSConv::wxCSConv(wxFontEncoding encoding
)
2780 if ( encoding
== wxFONTENCODING_MAX
|| encoding
== wxFONTENCODING_DEFAULT
)
2782 wxFAIL_MSG( _T("invalid encoding value in wxCSConv ctor") );
2784 encoding
= wxFONTENCODING_SYSTEM
;
2789 m_encoding
= encoding
;
2792 wxCSConv::~wxCSConv()
2797 wxCSConv::wxCSConv(const wxCSConv
& conv
)
2802 SetName(conv
.m_name
);
2803 m_encoding
= conv
.m_encoding
;
2806 wxCSConv
& wxCSConv::operator=(const wxCSConv
& conv
)
2810 SetName(conv
.m_name
);
2811 m_encoding
= conv
.m_encoding
;
2816 void wxCSConv::Clear()
2825 void wxCSConv::SetName(const wxChar
*charset
)
2829 m_name
= wxStrdup(charset
);
2835 #include "wx/hashmap.h"
2837 WX_DECLARE_HASH_MAP( wxFontEncoding
, wxString
, wxIntegerHash
, wxIntegerEqual
,
2838 wxEncodingNameCache
);
2840 static wxEncodingNameCache gs_nameCache
;
2843 wxMBConv
*wxCSConv::DoCreate() const
2846 wxLogTrace(TRACE_STRCONV
,
2847 wxT("creating conversion for %s"),
2849 : wxFontMapperBase::GetEncodingName(m_encoding
).c_str()));
2850 #endif // wxUSE_FONTMAP
2852 // check for the special case of ASCII or ISO8859-1 charset: as we have
2853 // special knowledge of it anyhow, we don't need to create a special
2854 // conversion object
2855 if ( m_encoding
== wxFONTENCODING_ISO8859_1
||
2856 m_encoding
== wxFONTENCODING_DEFAULT
)
2858 // don't convert at all
2862 // we trust OS to do conversion better than we can so try external
2863 // conversion methods first
2865 // the full order is:
2866 // 1. OS conversion (iconv() under Unix or Win32 API)
2867 // 2. hard coded conversions for UTF
2868 // 3. wxEncodingConverter as fall back
2874 #endif // !wxUSE_FONTMAP
2876 wxString
name(m_name
);
2877 wxFontEncoding
encoding(m_encoding
);
2879 if ( !name
.empty() )
2881 wxMBConv_iconv
*conv
= new wxMBConv_iconv(name
);
2889 wxFontMapperBase::Get()->CharsetToEncoding(name
, false);
2890 #endif // wxUSE_FONTMAP
2894 const wxEncodingNameCache::iterator it
= gs_nameCache
.find(encoding
);
2895 if ( it
!= gs_nameCache
.end() )
2897 if ( it
->second
.empty() )
2900 wxMBConv_iconv
*conv
= new wxMBConv_iconv(it
->second
);
2907 const wxChar
** names
= wxFontMapperBase::GetAllEncodingNames(encoding
);
2909 for ( ; *names
; ++names
)
2911 wxMBConv_iconv
*conv
= new wxMBConv_iconv(*names
);
2914 gs_nameCache
[encoding
] = *names
;
2921 gs_nameCache
[encoding
] = _T(""); // cache the failure
2923 #endif // wxUSE_FONTMAP
2925 #endif // HAVE_ICONV
2927 #ifdef wxHAVE_WIN32_MB2WC
2930 wxMBConv_win32
*conv
= m_name
? new wxMBConv_win32(m_name
)
2931 : new wxMBConv_win32(m_encoding
);
2940 #endif // wxHAVE_WIN32_MB2WC
2941 #if defined(__WXMAC__)
2943 // leave UTF16 and UTF32 to the built-ins of wx
2944 if ( m_name
|| ( m_encoding
< wxFONTENCODING_UTF16BE
||
2945 ( m_encoding
>= wxFONTENCODING_MACMIN
&& m_encoding
<= wxFONTENCODING_MACMAX
) ) )
2949 wxMBConv_mac
*conv
= m_name
? new wxMBConv_mac(m_name
)
2950 : new wxMBConv_mac(m_encoding
);
2952 wxMBConv_mac
*conv
= new wxMBConv_mac(m_encoding
);
2961 #if defined(__WXCOCOA__)
2963 if ( m_name
|| ( m_encoding
<= wxFONTENCODING_UTF16
) )
2967 wxMBConv_cocoa
*conv
= m_name
? new wxMBConv_cocoa(m_name
)
2968 : new wxMBConv_cocoa(m_encoding
);
2970 wxMBConv_cocoa
*conv
= new wxMBConv_cocoa(m_encoding
);
2980 wxFontEncoding enc
= m_encoding
;
2982 if ( enc
== wxFONTENCODING_SYSTEM
&& m_name
)
2984 // use "false" to suppress interactive dialogs -- we can be called from
2985 // anywhere and popping up a dialog from here is the last thing we want to
2987 enc
= wxFontMapperBase::Get()->CharsetToEncoding(m_name
, false);
2989 #endif // wxUSE_FONTMAP
2993 case wxFONTENCODING_UTF7
:
2994 return new wxMBConvUTF7
;
2996 case wxFONTENCODING_UTF8
:
2997 return new wxMBConvUTF8
;
2999 case wxFONTENCODING_UTF16BE
:
3000 return new wxMBConvUTF16BE
;
3002 case wxFONTENCODING_UTF16LE
:
3003 return new wxMBConvUTF16LE
;
3005 case wxFONTENCODING_UTF32BE
:
3006 return new wxMBConvUTF32BE
;
3008 case wxFONTENCODING_UTF32LE
:
3009 return new wxMBConvUTF32LE
;
3012 // nothing to do but put here to suppress gcc warnings
3019 wxMBConv_wxwin
*conv
= m_name
? new wxMBConv_wxwin(m_name
)
3020 : new wxMBConv_wxwin(m_encoding
);
3026 #endif // wxUSE_FONTMAP
3028 // NB: This is a hack to prevent deadlock. What could otherwise happen
3029 // in Unicode build: wxConvLocal creation ends up being here
3030 // because of some failure and logs the error. But wxLog will try to
3031 // attach timestamp, for which it will need wxConvLocal (to convert
3032 // time to char* and then wchar_t*), but that fails, tries to log
3033 // error, but wxLog has a (already locked) critical section that
3034 // guards static buffer.
3035 static bool alreadyLoggingError
= false;
3036 if (!alreadyLoggingError
)
3038 alreadyLoggingError
= true;
3039 wxLogError(_("Cannot convert from the charset '%s'!"),
3043 wxFontMapperBase::GetEncodingDescription(m_encoding
).c_str()
3044 #else // !wxUSE_FONTMAP
3045 wxString::Format(_("encoding %s"), m_encoding
).c_str()
3046 #endif // wxUSE_FONTMAP/!wxUSE_FONTMAP
3048 alreadyLoggingError
= false;
3054 void wxCSConv::CreateConvIfNeeded() const
3058 wxCSConv
*self
= (wxCSConv
*)this; // const_cast
3061 // if we don't have neither the name nor the encoding, use the default
3062 // encoding for this system
3063 if ( !m_name
&& m_encoding
== wxFONTENCODING_SYSTEM
)
3065 self
->m_name
= wxStrdup(wxLocale::GetSystemEncodingName());
3067 #endif // wxUSE_INTL
3069 self
->m_convReal
= DoCreate();
3070 self
->m_deferred
= false;
3074 size_t wxCSConv::MB2WC(wchar_t *buf
, const char *psz
, size_t n
) const
3076 CreateConvIfNeeded();
3079 return m_convReal
->MB2WC(buf
, psz
, n
);
3082 size_t len
= strlen(psz
);
3086 for (size_t c
= 0; c
<= len
; c
++)
3087 buf
[c
] = (unsigned char)(psz
[c
]);
3093 size_t wxCSConv::WC2MB(char *buf
, const wchar_t *psz
, size_t n
) const
3095 CreateConvIfNeeded();
3098 return m_convReal
->WC2MB(buf
, psz
, n
);
3101 const size_t len
= wxWcslen(psz
);
3104 for (size_t c
= 0; c
<= len
; c
++)
3108 buf
[c
] = (char)psz
[c
];
3113 for (size_t c
= 0; c
<= len
; c
++)
3123 size_t wxCSConv::GetMBNulLen() const
3125 CreateConvIfNeeded();
3129 return m_convReal
->GetMBNulLen();
3135 // ----------------------------------------------------------------------------
3137 // ----------------------------------------------------------------------------
3140 static wxMBConv_win32 wxConvLibcObj
;
3141 #elif defined(__WXMAC__) && !defined(__MACH__)
3142 static wxMBConv_mac wxConvLibcObj
;
3144 static wxMBConvLibc wxConvLibcObj
;
3147 static wxCSConv
wxConvLocalObj(wxFONTENCODING_SYSTEM
);
3148 static wxCSConv
wxConvISO8859_1Obj(wxFONTENCODING_ISO8859_1
);
3149 static wxMBConvUTF7 wxConvUTF7Obj
;
3150 static wxMBConvUTF8 wxConvUTF8Obj
;
3152 WXDLLIMPEXP_DATA_BASE(wxMBConv
&) wxConvLibc
= wxConvLibcObj
;
3153 WXDLLIMPEXP_DATA_BASE(wxCSConv
&) wxConvLocal
= wxConvLocalObj
;
3154 WXDLLIMPEXP_DATA_BASE(wxCSConv
&) wxConvISO8859_1
= wxConvISO8859_1Obj
;
3155 WXDLLIMPEXP_DATA_BASE(wxMBConvUTF7
&) wxConvUTF7
= wxConvUTF7Obj
;
3156 WXDLLIMPEXP_DATA_BASE(wxMBConvUTF8
&) wxConvUTF8
= wxConvUTF8Obj
;
3157 WXDLLIMPEXP_DATA_BASE(wxMBConv
*) wxConvCurrent
= &wxConvLibcObj
;
3158 WXDLLIMPEXP_DATA_BASE(wxMBConv
*) wxConvFileName
= &
3166 #else // !wxUSE_WCHAR_T
3168 // stand-ins in absence of wchar_t
3169 WXDLLIMPEXP_DATA_BASE(wxMBConv
) wxConvLibc
,
3174 #endif // wxUSE_WCHAR_T/!wxUSE_WCHAR_T