]> git.saurik.com Git - wxWidgets.git/blame_incremental - src/common/strconv.cpp
projects / wxWidgets.git / blame_incremental
? search:
summary | shortlog | log | commit | commitdiff | tree
blob | blame (non-incremental) | history | HEAD
cvs to svn
[wxWidgets.git] / src / common / strconv.cpp
... / ...
CommitLineData
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)
6// Modified by:
7// Created: 29/01/98
8// RCS-ID: $Id$
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/////////////////////////////////////////////////////////////////////////////
14
15// For compilers that support precompilation, includes "wx.h".
16#include "wx/wxprec.h"
17
18#ifdef __BORLANDC__
19 #pragma hdrstop
20#endif //__BORLANDC__
21
22#ifndef WX_PRECOMP
23 #include "wx/intl.h"
24 #include "wx/log.h"
25 #include "wx/utils.h"
26 #include "wx/hashmap.h"
27#endif
28
29#include "wx/strconv.h"
30
31#if wxUSE_WCHAR_T
32
33#ifndef __WXWINCE__
34#include <errno.h>
35#endif
36
37#include <ctype.h>
38#include <string.h>
39#include <stdlib.h>
40
41#if defined(__WIN32__) && !defined(__WXMICROWIN__)
42 #include "wx/msw/private.h"
43 #include "wx/msw/missing.h"
44 #define wxHAVE_WIN32_MB2WC
45#endif
46
47#ifdef __SALFORDC__
48 #include <clib.h>
49#endif
50
51#ifdef HAVE_ICONV
52 #include <iconv.h>
53 #include "wx/thread.h"
54#endif
55
56#include "wx/encconv.h"
57#include "wx/fontmap.h"
58
59#ifdef __WXMAC__
60#ifndef __DARWIN__
61#include <ATSUnicode.h>
62#include <TextCommon.h>
63#include <TextEncodingConverter.h>
64#endif
65
66// includes Mac headers
67#include "wx/mac/private.h"
68#endif
69
70
71#define TRACE_STRCONV _T("strconv")
72
73// WC_UTF16 is defined only if sizeof(wchar_t) == 2, otherwise it's supposed to
74// be 4 bytes
75#if SIZEOF_WCHAR_T == 2
76 #define WC_UTF16
77#endif
78
79
80// ============================================================================
81// implementation
82// ============================================================================
83
84// helper function of cMB2WC(): check if n bytes at this location are all NUL
85static bool NotAllNULs(const char *p, size_t n)
86{
87 while ( n && *p++ == '\0' )
88 n--;
89
90 return n != 0;
91}
92
93// ----------------------------------------------------------------------------
94// UTF-16 en/decoding to/from UCS-4 with surrogates handling
95// ----------------------------------------------------------------------------
96
97static size_t encode_utf16(wxUint32 input, wxUint16 *output)
98{
99 if (input <= 0xffff)
100 {
101 if (output)
102 *output = (wxUint16) input;
103
104 return 1;
105 }
106 else if (input >= 0x110000)
107 {
108 return wxCONV_FAILED;
109 }
110 else
111 {
112 if (output)
113 {
114 *output++ = (wxUint16) ((input >> 10) + 0xd7c0);
115 *output = (wxUint16) ((input & 0x3ff) + 0xdc00);
116 }
117
118 return 2;
119 }
120}
121
122static size_t decode_utf16(const wxUint16* input, wxUint32& output)
123{
124 if ((*input < 0xd800) || (*input > 0xdfff))
125 {
126 output = *input;
127 return 1;
128 }
129 else if ((input[1] < 0xdc00) || (input[1] > 0xdfff))
130 {
131 output = *input;
132 return wxCONV_FAILED;
133 }
134 else
135 {
136 output = ((input[0] - 0xd7c0) << 10) + (input[1] - 0xdc00);
137 return 2;
138 }
139}
140
141#ifdef WC_UTF16
142 typedef wchar_t wxDecodeSurrogate_t;
143#else // !WC_UTF16
144 typedef wxUint16 wxDecodeSurrogate_t;
145#endif // WC_UTF16/!WC_UTF16
146
147// returns the next UTF-32 character from the wchar_t buffer and advances the
148// pointer to the character after this one
149//
150// if an invalid character is found, *pSrc is set to NULL, the caller must
151// check for this
152static wxUint32 wxDecodeSurrogate(const wxDecodeSurrogate_t **pSrc)
153{
154 wxUint32 out;
155 const size_t
156 n = decode_utf16(wx_reinterpret_cast(const wxUint16 *, *pSrc), out);
157 if ( n == wxCONV_FAILED )
158 *pSrc = NULL;
159 else
160 *pSrc += n;
161
162 return out;
163}
164
165// ----------------------------------------------------------------------------
166// wxMBConv
167// ----------------------------------------------------------------------------
168
169size_t
170wxMBConv::ToWChar(wchar_t *dst, size_t dstLen,
171 const char *src, size_t srcLen) const
172{
173 // although new conversion classes are supposed to implement this function
174 // directly, the existins ones only implement the old MB2WC() and so, to
175 // avoid to have to rewrite all conversion classes at once, we provide a
176 // default (but not efficient) implementation of this one in terms of the
177 // old function by copying the input to ensure that it's NUL-terminated and
178 // then using MB2WC() to convert it
179
180 // the number of chars [which would be] written to dst [if it were not NULL]
181 size_t dstWritten = 0;
182
183 // the number of NULs terminating this string
184 size_t nulLen = 0; // not really needed, but just to avoid warnings
185
186 // if we were not given the input size we just have to assume that the
187 // string is properly terminated as we have no way of knowing how long it
188 // is anyhow, but if we do have the size check whether there are enough
189 // NULs at the end
190 wxCharBuffer bufTmp;
191 const char *srcEnd;
192 if ( srcLen != wxNO_LEN )
193 {
194 // we need to know how to find the end of this string
195 nulLen = GetMBNulLen();
196 if ( nulLen == wxCONV_FAILED )
197 return wxCONV_FAILED;
198
199 // if there are enough NULs we can avoid the copy
200 if ( srcLen < nulLen || NotAllNULs(src + srcLen - nulLen, nulLen) )
201 {
202 // make a copy in order to properly NUL-terminate the string
203 bufTmp = wxCharBuffer(srcLen + nulLen - 1 /* 1 will be added */);
204 char * const p = bufTmp.data();
205 memcpy(p, src, srcLen);
206 for ( char *s = p + srcLen; s < p + srcLen + nulLen; s++ )
207 *s = '\0';
208
209 src = bufTmp;
210 }
211
212 srcEnd = src + srcLen;
213 }
214 else // quit after the first loop iteration
215 {
216 srcEnd = NULL;
217 }
218
219 for ( ;; )
220 {
221 // try to convert the current chunk
222 size_t lenChunk = MB2WC(NULL, src, 0);
223 if ( lenChunk == wxCONV_FAILED )
224 return wxCONV_FAILED;
225
226 lenChunk++; // for the L'\0' at the end of this chunk
227
228 dstWritten += lenChunk;
229
230 if ( lenChunk == 1 )
231 {
232 // nothing left in the input string, conversion succeeded
233 break;
234 }
235
236 if ( dst )
237 {
238 if ( dstWritten > dstLen )
239 return wxCONV_FAILED;
240
241 if ( MB2WC(dst, src, lenChunk) == wxCONV_FAILED )
242 return wxCONV_FAILED;
243
244 dst += lenChunk;
245 }
246
247 if ( !srcEnd )
248 {
249 // we convert just one chunk in this case as this is the entire
250 // string anyhow
251 break;
252 }
253
254 // advance the input pointer past the end of this chunk
255 while ( NotAllNULs(src, nulLen) )
256 {
257 // notice that we must skip over multiple bytes here as we suppose
258 // that if NUL takes 2 or 4 bytes, then all the other characters do
259 // too and so if advanced by a single byte we might erroneously
260 // detect sequences of NUL bytes in the middle of the input
261 src += nulLen;
262 }
263
264 src += nulLen; // skipping over its terminator as well
265
266 // note that ">=" (and not just "==") is needed here as the terminator
267 // we skipped just above could be inside or just after the buffer
268 // delimited by inEnd
269 if ( src >= srcEnd )
270 break;
271 }
272
273 return dstWritten;
274}
275
276size_t
277wxMBConv::FromWChar(char *dst, size_t dstLen,
278 const wchar_t *src, size_t srcLen) const
279{
280 // the number of chars [which would be] written to dst [if it were not NULL]
281 size_t dstWritten = 0;
282
283 // make a copy of the input string unless it is already properly
284 // NUL-terminated
285 //
286 // if we don't know its length we have no choice but to assume that it is,
287 // indeed, properly terminated
288 wxWCharBuffer bufTmp;
289 if ( srcLen == wxNO_LEN )
290 {
291 srcLen = wxWcslen(src) + 1;
292 }
293 else if ( srcLen != 0 && src[srcLen - 1] != L'\0' )
294 {
295 // make a copy in order to properly NUL-terminate the string
296 bufTmp = wxWCharBuffer(srcLen);
297 memcpy(bufTmp.data(), src, srcLen * sizeof(wchar_t));
298 src = bufTmp;
299 }
300
301 const size_t lenNul = GetMBNulLen();
302 for ( const wchar_t * const srcEnd = src + srcLen;
303 src < srcEnd;
304 src += wxWcslen(src) + 1 /* skip L'\0' too */ )
305 {
306 // try to convert the current chunk
307 size_t lenChunk = WC2MB(NULL, src, 0);
308
309 if ( lenChunk == wxCONV_FAILED )
310 return wxCONV_FAILED;
311
312 lenChunk += lenNul;
313 dstWritten += lenChunk;
314
315 if ( dst )
316 {
317 if ( dstWritten > dstLen )
318 return wxCONV_FAILED;
319
320 if ( WC2MB(dst, src, lenChunk) == wxCONV_FAILED )
321 return wxCONV_FAILED;
322
323 dst += lenChunk;
324 }
325 }
326
327 return dstWritten;
328}
329
330size_t wxMBConv::MB2WC(wchar_t *outBuff, const char *inBuff, size_t outLen) const
331{
332 size_t rc = ToWChar(outBuff, outLen, inBuff);
333 if ( rc != wxCONV_FAILED )
334 {
335 // ToWChar() returns the buffer length, i.e. including the trailing
336 // NUL, while this method doesn't take it into account
337 rc--;
338 }
339
340 return rc;
341}
342
343size_t wxMBConv::WC2MB(char *outBuff, const wchar_t *inBuff, size_t outLen) const
344{
345 size_t rc = FromWChar(outBuff, outLen, inBuff);
346 if ( rc != wxCONV_FAILED )
347 {
348 rc -= GetMBNulLen();
349 }
350
351 return rc;
352}
353
354wxMBConv::~wxMBConv()
355{
356 // nothing to do here (necessary for Darwin linking probably)
357}
358
359const wxWCharBuffer wxMBConv::cMB2WC(const char *psz) const
360{
361 if ( psz )
362 {
363 // calculate the length of the buffer needed first
364 const size_t nLen = MB2WC(NULL, psz, 0);
365 if ( nLen != wxCONV_FAILED )
366 {
367 // now do the actual conversion
368 wxWCharBuffer buf(nLen /* +1 added implicitly */);
369
370 // +1 for the trailing NULL
371 if ( MB2WC(buf.data(), psz, nLen + 1) != wxCONV_FAILED )
372 return buf;
373 }
374 }
375
376 return wxWCharBuffer();
377}
378
379const wxCharBuffer wxMBConv::cWC2MB(const wchar_t *pwz) const
380{
381 if ( pwz )
382 {
383 const size_t nLen = WC2MB(NULL, pwz, 0);
384 if ( nLen != wxCONV_FAILED )
385 {
386 // extra space for trailing NUL(s)
387 static const size_t extraLen = GetMaxMBNulLen();
388
389 wxCharBuffer buf(nLen + extraLen - 1);
390 if ( WC2MB(buf.data(), pwz, nLen + extraLen) != wxCONV_FAILED )
391 return buf;
392 }
393 }
394
395 return wxCharBuffer();
396}
397
398const wxWCharBuffer
399wxMBConv::cMB2WC(const char *inBuff, size_t inLen, size_t *outLen) const
400{
401 const size_t dstLen = ToWChar(NULL, 0, inBuff, inLen);
402 if ( dstLen != wxCONV_FAILED )
403 {
404 wxWCharBuffer wbuf(dstLen - 1);
405 if ( ToWChar(wbuf.data(), dstLen, inBuff, inLen) != wxCONV_FAILED )
406 {
407 if ( outLen )
408 {
409 *outLen = dstLen;
410 if ( wbuf[dstLen - 1] == L'\0' )
411 (*outLen)--;
412 }
413
414 return wbuf;
415 }
416 }
417
418 if ( outLen )
419 *outLen = 0;
420
421 return wxWCharBuffer();
422}
423
424const wxCharBuffer
425wxMBConv::cWC2MB(const wchar_t *inBuff, size_t inLen, size_t *outLen) const
426{
427 size_t dstLen = FromWChar(NULL, 0, inBuff, inLen);
428 if ( dstLen != wxCONV_FAILED )
429 {
430 // special case of empty input: can't allocate 0 size buffer below as
431 // wxCharBuffer insists on NUL-terminating it
432 wxCharBuffer buf(dstLen ? dstLen - 1 : 1);
433 if ( FromWChar(buf.data(), dstLen, inBuff, inLen) != wxCONV_FAILED )
434 {
435 if ( outLen )
436 {
437 *outLen = dstLen;
438
439 const size_t nulLen = GetMBNulLen();
440 if ( dstLen >= nulLen &&
441 !NotAllNULs(buf.data() + dstLen - nulLen, nulLen) )
442 {
443 // in this case the output is NUL-terminated and we're not
444 // supposed to count NUL
445 *outLen -= nulLen;
446 }
447 }
448
449 return buf;
450 }
451 }
452
453 if ( outLen )
454 *outLen = 0;
455
456 return wxCharBuffer();
457}
458
459// ----------------------------------------------------------------------------
460// wxMBConvLibc
461// ----------------------------------------------------------------------------
462
463size_t wxMBConvLibc::MB2WC(wchar_t *buf, const char *psz, size_t n) const
464{
465 return wxMB2WC(buf, psz, n);
466}
467
468size_t wxMBConvLibc::WC2MB(char *buf, const wchar_t *psz, size_t n) const
469{
470 return wxWC2MB(buf, psz, n);
471}
472
473// ----------------------------------------------------------------------------
474// wxConvBrokenFileNames
475// ----------------------------------------------------------------------------
476
477#ifdef __UNIX__
478
479wxConvBrokenFileNames::wxConvBrokenFileNames(const wxString& charset)
480{
481 if ( wxStricmp(charset, _T("UTF-8")) == 0 ||
482 wxStricmp(charset, _T("UTF8")) == 0 )
483 m_conv = new wxMBConvUTF8(wxMBConvUTF8::MAP_INVALID_UTF8_TO_PUA);
484 else
485 m_conv = new wxCSConv(charset);
486}
487
488#endif // __UNIX__
489
490// ----------------------------------------------------------------------------
491// UTF-7
492// ----------------------------------------------------------------------------
493
494// Implementation (C) 2004 Fredrik Roubert
495
496//
497// BASE64 decoding table
498//
499static const unsigned char utf7unb64[] =
500{
501 0xff, 0xff, 0xff, 0xff, 0xff, 0xff, 0xff, 0xff,
502 0xff, 0xff, 0xff, 0xff, 0xff, 0xff, 0xff, 0xff,
503 0xff, 0xff, 0xff, 0xff, 0xff, 0xff, 0xff, 0xff,
504 0xff, 0xff, 0xff, 0xff, 0xff, 0xff, 0xff, 0xff,
505 0xff, 0xff, 0xff, 0xff, 0xff, 0xff, 0xff, 0xff,
506 0xff, 0xff, 0xff, 0x3e, 0xff, 0xff, 0xff, 0x3f,
507 0x34, 0x35, 0x36, 0x37, 0x38, 0x39, 0x3a, 0x3b,
508 0x3c, 0x3d, 0xff, 0xff, 0xff, 0xff, 0xff, 0xff,
509 0xff, 0x00, 0x01, 0x02, 0x03, 0x04, 0x05, 0x06,
510 0x07, 0x08, 0x09, 0x0a, 0x0b, 0x0c, 0x0d, 0x0e,
511 0x0f, 0x10, 0x11, 0x12, 0x13, 0x14, 0x15, 0x16,
512 0x17, 0x18, 0x19, 0xff, 0xff, 0xff, 0xff, 0xff,
513 0xff, 0x1a, 0x1b, 0x1c, 0x1d, 0x1e, 0x1f, 0x20,
514 0x21, 0x22, 0x23, 0x24, 0x25, 0x26, 0x27, 0x28,
515 0x29, 0x2a, 0x2b, 0x2c, 0x2d, 0x2e, 0x2f, 0x30,
516 0x31, 0x32, 0x33, 0xff, 0xff, 0xff, 0xff, 0xff,
517 0xff, 0xff, 0xff, 0xff, 0xff, 0xff, 0xff, 0xff,
518 0xff, 0xff, 0xff, 0xff, 0xff, 0xff, 0xff, 0xff,
519 0xff, 0xff, 0xff, 0xff, 0xff, 0xff, 0xff, 0xff,
520 0xff, 0xff, 0xff, 0xff, 0xff, 0xff, 0xff, 0xff,
521 0xff, 0xff, 0xff, 0xff, 0xff, 0xff, 0xff, 0xff,
522 0xff, 0xff, 0xff, 0xff, 0xff, 0xff, 0xff, 0xff,
523 0xff, 0xff, 0xff, 0xff, 0xff, 0xff, 0xff, 0xff,
524 0xff, 0xff, 0xff, 0xff, 0xff, 0xff, 0xff, 0xff,
525 0xff, 0xff, 0xff, 0xff, 0xff, 0xff, 0xff, 0xff,
526 0xff, 0xff, 0xff, 0xff, 0xff, 0xff, 0xff, 0xff,
527 0xff, 0xff, 0xff, 0xff, 0xff, 0xff, 0xff, 0xff,
528 0xff, 0xff, 0xff, 0xff, 0xff, 0xff, 0xff, 0xff,
529 0xff, 0xff, 0xff, 0xff, 0xff, 0xff, 0xff, 0xff,
530 0xff, 0xff, 0xff, 0xff, 0xff, 0xff, 0xff, 0xff,
531 0xff, 0xff, 0xff, 0xff, 0xff, 0xff, 0xff, 0xff,
532 0xff, 0xff, 0xff, 0xff, 0xff, 0xff, 0xff, 0xff
533};
534
535size_t wxMBConvUTF7::MB2WC(wchar_t *buf, const char *psz, size_t n) const
536{
537 size_t len = 0;
538
539 while ( *psz && (!buf || (len < n)) )
540 {
541 unsigned char cc = *psz++;
542 if (cc != '+')
543 {
544 // plain ASCII char
545 if (buf)
546 *buf++ = cc;
547 len++;
548 }
549 else if (*psz == '-')
550 {
551 // encoded plus sign
552 if (buf)
553 *buf++ = cc;
554 len++;
555 psz++;
556 }
557 else // start of BASE64 encoded string
558 {
559 bool lsb, ok;
560 unsigned int d, l;
561 for ( ok = lsb = false, d = 0, l = 0;
562 (cc = utf7unb64[(unsigned char)*psz]) != 0xff;
563 psz++ )
564 {
565 d <<= 6;
566 d += cc;
567 for (l += 6; l >= 8; lsb = !lsb)
568 {
569 unsigned char c = (unsigned char)((d >> (l -= 8)) % 256);
570 if (lsb)
571 {
572 if (buf)
573 *buf++ |= c;
574 len ++;
575 }
576 else
577 {
578 if (buf)
579 *buf = (wchar_t)(c << 8);
580 }
581
582 ok = true;
583 }
584 }
585
586 if ( !ok )
587 {
588 // in valid UTF7 we should have valid characters after '+'
589 return wxCONV_FAILED;
590 }
591
592 if (*psz == '-')
593 psz++;
594 }
595 }
596
597 if ( buf && (len < n) )
598 *buf = '\0';
599
600 return len;
601}
602
603//
604// BASE64 encoding table
605//
606static const unsigned char utf7enb64[] =
607{
608 'A', 'B', 'C', 'D', 'E', 'F', 'G', 'H',
609 'I', 'J', 'K', 'L', 'M', 'N', 'O', 'P',
610 'Q', 'R', 'S', 'T', 'U', 'V', 'W', 'X',
611 'Y', 'Z', 'a', 'b', 'c', 'd', 'e', 'f',
612 'g', 'h', 'i', 'j', 'k', 'l', 'm', 'n',
613 'o', 'p', 'q', 'r', 's', 't', 'u', 'v',
614 'w', 'x', 'y', 'z', '0', '1', '2', '3',
615 '4', '5', '6', '7', '8', '9', '+', '/'
616};
617
618//
619// UTF-7 encoding table
620//
621// 0 - Set D (directly encoded characters)
622// 1 - Set O (optional direct characters)
623// 2 - whitespace characters (optional)
624// 3 - special characters
625//
626static const unsigned char utf7encode[128] =
627{
628 3, 3, 3, 3, 3, 3, 3, 3, 3, 2, 2, 3, 3, 2, 3, 3,
629 3, 3, 3, 3, 3, 3, 3, 3, 3, 3, 3, 3, 3, 3, 3, 3,
630 2, 1, 1, 1, 1, 1, 1, 0, 0, 0, 1, 3, 0, 0, 0, 3,
631 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 1, 1, 1, 1, 0,
632 1, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0,
633 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 1, 3, 1, 1, 1,
634 1, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0,
635 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 1, 1, 1, 3, 3
636};
637
638size_t wxMBConvUTF7::WC2MB(char *buf, const wchar_t *psz, size_t n) const
639{
640 size_t len = 0;
641
642 while (*psz && ((!buf) || (len < n)))
643 {
644 wchar_t cc = *psz++;
645 if (cc < 0x80 && utf7encode[cc] < 1)
646 {
647 // plain ASCII char
648 if (buf)
649 *buf++ = (char)cc;
650
651 len++;
652 }
653#ifndef WC_UTF16
654 else if (((wxUint32)cc) > 0xffff)
655 {
656 // no surrogate pair generation (yet?)
657 return wxCONV_FAILED;
658 }
659#endif
660 else
661 {
662 if (buf)
663 *buf++ = '+';
664
665 len++;
666 if (cc != '+')
667 {
668 // BASE64 encode string
669 unsigned int lsb, d, l;
670 for (d = 0, l = 0; /*nothing*/; psz++)
671 {
672 for (lsb = 0; lsb < 2; lsb ++)
673 {
674 d <<= 8;
675 d += lsb ? cc & 0xff : (cc & 0xff00) >> 8;
676
677 for (l += 8; l >= 6; )
678 {
679 l -= 6;
680 if (buf)
681 *buf++ = utf7enb64[(d >> l) % 64];
682 len++;
683 }
684 }
685
686 cc = *psz;
687 if (!(cc) || (cc < 0x80 && utf7encode[cc] < 1))
688 break;
689 }
690
691 if (l != 0)
692 {
693 if (buf)
694 *buf++ = utf7enb64[((d % 16) << (6 - l)) % 64];
695
696 len++;
697 }
698 }
699
700 if (buf)
701 *buf++ = '-';
702 len++;
703 }
704 }
705
706 if (buf && (len < n))
707 *buf = 0;
708
709 return len;
710}
711
712// ----------------------------------------------------------------------------
713// UTF-8
714// ----------------------------------------------------------------------------
715
716static wxUint32 utf8_max[]=
717 { 0x7f, 0x7ff, 0xffff, 0x1fffff, 0x3ffffff, 0x7fffffff, 0xffffffff };
718
719// boundaries of the private use area we use to (temporarily) remap invalid
720// characters invalid in a UTF-8 encoded string
721const wxUint32 wxUnicodePUA = 0x100000;
722const wxUint32 wxUnicodePUAEnd = wxUnicodePUA + 256;
723
724size_t wxMBConvUTF8::MB2WC(wchar_t *buf, const char *psz, size_t n) const
725{
726 size_t len = 0;
727
728 while (*psz && ((!buf) || (len < n)))
729 {
730 const char *opsz = psz;
731 bool invalid = false;
732 unsigned char cc = *psz++, fc = cc;
733 unsigned cnt;
734 for (cnt = 0; fc & 0x80; cnt++)
735 fc <<= 1;
736
737 if (!cnt)
738 {
739 // plain ASCII char
740 if (buf)
741 *buf++ = cc;
742 len++;
743
744 // escape the escape character for octal escapes
745 if ((m_options & MAP_INVALID_UTF8_TO_OCTAL)
746 && cc == '\\' && (!buf || len < n))
747 {
748 if (buf)
749 *buf++ = cc;
750 len++;
751 }
752 }
753 else
754 {
755 cnt--;
756 if (!cnt)
757 {
758 // invalid UTF-8 sequence
759 invalid = true;
760 }
761 else
762 {
763 unsigned ocnt = cnt - 1;
764 wxUint32 res = cc & (0x3f >> cnt);
765 while (cnt--)
766 {
767 cc = *psz;
768 if ((cc & 0xC0) != 0x80)
769 {
770 // invalid UTF-8 sequence
771 invalid = true;
772 break;
773 }
774
775 psz++;
776 res = (res << 6) | (cc & 0x3f);
777 }
778
779 if (invalid || res <= utf8_max[ocnt])
780 {
781 // illegal UTF-8 encoding
782 invalid = true;
783 }
784 else if ((m_options & MAP_INVALID_UTF8_TO_PUA) &&
785 res >= wxUnicodePUA && res < wxUnicodePUAEnd)
786 {
787 // if one of our PUA characters turns up externally
788 // it must also be treated as an illegal sequence
789 // (a bit like you have to escape an escape character)
790 invalid = true;
791 }
792 else
793 {
794#ifdef WC_UTF16
795 // cast is ok because wchar_t == wxUuint16 if WC_UTF16
796 size_t pa = encode_utf16(res, (wxUint16 *)buf);
797 if (pa == wxCONV_FAILED)
798 {
799 invalid = true;
800 }
801 else
802 {
803 if (buf)
804 buf += pa;
805 len += pa;
806 }
807#else // !WC_UTF16
808 if (buf)
809 *buf++ = (wchar_t)res;
810 len++;
811#endif // WC_UTF16/!WC_UTF16
812 }
813 }
814
815 if (invalid)
816 {
817 if (m_options & MAP_INVALID_UTF8_TO_PUA)
818 {
819 while (opsz < psz && (!buf || len < n))
820 {
821#ifdef WC_UTF16
822 // cast is ok because wchar_t == wxUuint16 if WC_UTF16
823 size_t pa = encode_utf16((unsigned char)*opsz + wxUnicodePUA, (wxUint16 *)buf);
824 wxASSERT(pa != wxCONV_FAILED);
825 if (buf)
826 buf += pa;
827 opsz++;
828 len += pa;
829#else
830 if (buf)
831 *buf++ = (wchar_t)(wxUnicodePUA + (unsigned char)*opsz);
832 opsz++;
833 len++;
834#endif
835 }
836 }
837 else if (m_options & MAP_INVALID_UTF8_TO_OCTAL)
838 {
839 while (opsz < psz && (!buf || len < n))
840 {
841 if ( buf && len + 3 < n )
842 {
843 unsigned char on = *opsz;
844 *buf++ = L'\\';
845 *buf++ = (wchar_t)( L'0' + on / 0100 );
846 *buf++ = (wchar_t)( L'0' + (on % 0100) / 010 );
847 *buf++ = (wchar_t)( L'0' + on % 010 );
848 }
849
850 opsz++;
851 len += 4;
852 }
853 }
854 else // MAP_INVALID_UTF8_NOT
855 {
856 return wxCONV_FAILED;
857 }
858 }
859 }
860 }
861
862 if (buf && (len < n))
863 *buf = 0;
864
865 return len;
866}
867
868static inline bool isoctal(wchar_t wch)
869{
870 return L'0' <= wch && wch <= L'7';
871}
872
873size_t wxMBConvUTF8::WC2MB(char *buf, const wchar_t *psz, size_t n) const
874{
875 size_t len = 0;
876
877 while (*psz && ((!buf) || (len < n)))
878 {
879 wxUint32 cc;
880
881#ifdef WC_UTF16
882 // cast is ok for WC_UTF16
883 size_t pa = decode_utf16((const wxUint16 *)psz, cc);
884 psz += (pa == wxCONV_FAILED) ? 1 : pa;
885#else
886 cc = (*psz++) & 0x7fffffff;
887#endif
888
889 if ( (m_options & MAP_INVALID_UTF8_TO_PUA)
890 && cc >= wxUnicodePUA && cc < wxUnicodePUAEnd )
891 {
892 if (buf)
893 *buf++ = (char)(cc - wxUnicodePUA);
894 len++;
895 }
896 else if ( (m_options & MAP_INVALID_UTF8_TO_OCTAL)
897 && cc == L'\\' && psz[0] == L'\\' )
898 {
899 if (buf)
900 *buf++ = (char)cc;
901 psz++;
902 len++;
903 }
904 else if ( (m_options & MAP_INVALID_UTF8_TO_OCTAL) &&
905 cc == L'\\' &&
906 isoctal(psz[0]) && isoctal(psz[1]) && isoctal(psz[2]) )
907 {
908 if (buf)
909 {
910 *buf++ = (char) ((psz[0] - L'0') * 0100 +
911 (psz[1] - L'0') * 010 +
912 (psz[2] - L'0'));
913 }
914
915 psz += 3;
916 len++;
917 }
918 else
919 {
920 unsigned cnt;
921 for (cnt = 0; cc > utf8_max[cnt]; cnt++)
922 {
923 }
924
925 if (!cnt)
926 {
927 // plain ASCII char
928 if (buf)
929 *buf++ = (char) cc;
930 len++;
931 }
932 else
933 {
934 len += cnt + 1;
935 if (buf)
936 {
937 *buf++ = (char) ((-128 >> cnt) | ((cc >> (cnt * 6)) & (0x3f >> cnt)));
938 while (cnt--)
939 *buf++ = (char) (0x80 | ((cc >> (cnt * 6)) & 0x3f));
940 }
941 }
942 }
943 }
944
945 if (buf && (len < n))
946 *buf = 0;
947
948 return len;
949}
950
951// ============================================================================
952// UTF-16
953// ============================================================================
954
955#ifdef WORDS_BIGENDIAN
956 #define wxMBConvUTF16straight wxMBConvUTF16BE
957 #define wxMBConvUTF16swap wxMBConvUTF16LE
958#else
959 #define wxMBConvUTF16swap wxMBConvUTF16BE
960 #define wxMBConvUTF16straight wxMBConvUTF16LE
961#endif
962
963/* static */
964size_t wxMBConvUTF16Base::GetLength(const char *src, size_t srcLen)
965{
966 if ( srcLen == wxNO_LEN )
967 {
968 // count the number of bytes in input, including the trailing NULs
969 const wxUint16 *inBuff = wx_reinterpret_cast(const wxUint16 *, src);
970 for ( srcLen = 1; *inBuff++; srcLen++ )
971 ;
972
973 srcLen *= BYTES_PER_CHAR;
974 }
975 else // we already have the length
976 {
977 // we can only convert an entire number of UTF-16 characters
978 if ( srcLen % BYTES_PER_CHAR )
979 return wxCONV_FAILED;
980 }
981
982 return srcLen;
983}
984
985// case when in-memory representation is UTF-16 too
986#ifdef WC_UTF16
987
988// ----------------------------------------------------------------------------
989// conversions without endianness change
990// ----------------------------------------------------------------------------
991
992size_t
993wxMBConvUTF16straight::ToWChar(wchar_t *dst, size_t dstLen,
994 const char *src, size_t srcLen) const
995{
996 // set up the scene for using memcpy() (which is presumably more efficient
997 // than copying the bytes one by one)
998 srcLen = GetLength(src, srcLen);
999 if ( srcLen == wxNO_LEN )
1000 return wxCONV_FAILED;
1001
1002 const size_t inLen = srcLen / BYTES_PER_CHAR;
1003 if ( dst )
1004 {
1005 if ( dstLen < inLen )
1006 return wxCONV_FAILED;
1007
1008 memcpy(dst, src, srcLen);
1009 }
1010
1011 return inLen;
1012}
1013
1014size_t
1015wxMBConvUTF16straight::FromWChar(char *dst, size_t dstLen,
1016 const wchar_t *src, size_t srcLen) const
1017{
1018 if ( srcLen == wxNO_LEN )
1019 srcLen = wxWcslen(src) + 1;
1020
1021 srcLen *= BYTES_PER_CHAR;
1022
1023 if ( dst )
1024 {
1025 if ( dstLen < srcLen )
1026 return wxCONV_FAILED;
1027
1028 memcpy(dst, src, srcLen);
1029 }
1030
1031 return srcLen;
1032}
1033
1034// ----------------------------------------------------------------------------
1035// endian-reversing conversions
1036// ----------------------------------------------------------------------------
1037
1038size_t
1039wxMBConvUTF16swap::ToWChar(wchar_t *dst, size_t dstLen,
1040 const char *src, size_t srcLen) const
1041{
1042 srcLen = GetLength(src, srcLen);
1043 if ( srcLen == wxNO_LEN )
1044 return wxCONV_FAILED;
1045
1046 srcLen /= BYTES_PER_CHAR;
1047
1048 if ( dst )
1049 {
1050 if ( dstLen < srcLen )
1051 return wxCONV_FAILED;
1052
1053 const wxUint16 *inBuff = wx_reinterpret_cast(const wxUint16 *, src);
1054 for ( size_t n = 0; n < srcLen; n++, inBuff++ )
1055 {
1056 *dst++ = wxUINT16_SWAP_ALWAYS(*inBuff);
1057 }
1058 }
1059
1060 return srcLen;
1061}
1062
1063size_t
1064wxMBConvUTF16swap::FromWChar(char *dst, size_t dstLen,
1065 const wchar_t *src, size_t srcLen) const
1066{
1067 if ( srcLen == wxNO_LEN )
1068 srcLen = wxWcslen(src) + 1;
1069
1070 srcLen *= BYTES_PER_CHAR;
1071
1072 if ( dst )
1073 {
1074 if ( dstLen < srcLen )
1075 return wxCONV_FAILED;
1076
1077 wxUint16 *outBuff = wx_reinterpret_cast(wxUint16 *, dst);
1078 for ( size_t n = 0; n < srcLen; n += BYTES_PER_CHAR, src++ )
1079 {
1080 *outBuff++ = wxUINT16_SWAP_ALWAYS(*src);
1081 }
1082 }
1083
1084 return srcLen;
1085}
1086
1087#else // !WC_UTF16: wchar_t is UTF-32
1088
1089// ----------------------------------------------------------------------------
1090// conversions without endianness change
1091// ----------------------------------------------------------------------------
1092
1093size_t
1094wxMBConvUTF16straight::ToWChar(wchar_t *dst, size_t dstLen,
1095 const char *src, size_t srcLen) const
1096{
1097 srcLen = GetLength(src, srcLen);
1098 if ( srcLen == wxNO_LEN )
1099 return wxCONV_FAILED;
1100
1101 const size_t inLen = srcLen / BYTES_PER_CHAR;
1102 if ( !dst )
1103 {
1104 // optimization: return maximal space which could be needed for this
1105 // string even if the real size could be smaller if the buffer contains
1106 // any surrogates
1107 return inLen;
1108 }
1109
1110 size_t outLen = 0;
1111 const wxUint16 *inBuff = wx_reinterpret_cast(const wxUint16 *, src);
1112 for ( const wxUint16 * const inEnd = inBuff + inLen; inBuff < inEnd; )
1113 {
1114 const wxUint32 ch = wxDecodeSurrogate(&inBuff);
1115 if ( !inBuff )
1116 return wxCONV_FAILED;
1117
1118 if ( ++outLen > dstLen )
1119 return wxCONV_FAILED;
1120
1121 *dst++ = ch;
1122 }
1123
1124
1125 return outLen;
1126}
1127
1128size_t
1129wxMBConvUTF16straight::FromWChar(char *dst, size_t dstLen,
1130 const wchar_t *src, size_t srcLen) const
1131{
1132 if ( srcLen == wxNO_LEN )
1133 srcLen = wxWcslen(src) + 1;
1134
1135 size_t outLen = 0;
1136 wxUint16 *outBuff = wx_reinterpret_cast(wxUint16 *, dst);
1137 for ( size_t n = 0; n < srcLen; n++ )
1138 {
1139 wxUint16 cc[2];
1140 const size_t numChars = encode_utf16(*src++, cc);
1141 if ( numChars == wxCONV_FAILED )
1142 return wxCONV_FAILED;
1143
1144 outLen += numChars * BYTES_PER_CHAR;
1145 if ( outBuff )
1146 {
1147 if ( outLen > dstLen )
1148 return wxCONV_FAILED;
1149
1150 *outBuff++ = cc[0];
1151 if ( numChars == 2 )
1152 {
1153 // second character of a surrogate
1154 *outBuff++ = cc[1];
1155 }
1156 }
1157 }
1158
1159 return outLen;
1160}
1161
1162// ----------------------------------------------------------------------------
1163// endian-reversing conversions
1164// ----------------------------------------------------------------------------
1165
1166size_t
1167wxMBConvUTF16swap::ToWChar(wchar_t *dst, size_t dstLen,
1168 const char *src, size_t srcLen) const
1169{
1170 srcLen = GetLength(src, srcLen);
1171 if ( srcLen == wxNO_LEN )
1172 return wxCONV_FAILED;
1173
1174 const size_t inLen = srcLen / BYTES_PER_CHAR;
1175 if ( !dst )
1176 {
1177 // optimization: return maximal space which could be needed for this
1178 // string even if the real size could be smaller if the buffer contains
1179 // any surrogates
1180 return inLen;
1181 }
1182
1183 size_t outLen = 0;
1184 const wxUint16 *inBuff = wx_reinterpret_cast(const wxUint16 *, src);
1185 for ( const wxUint16 * const inEnd = inBuff + inLen; inBuff < inEnd; )
1186 {
1187 wxUint32 ch;
1188 wxUint16 tmp[2];
1189
1190 tmp[0] = wxUINT16_SWAP_ALWAYS(*inBuff);
1191 inBuff++;
1192 tmp[1] = wxUINT16_SWAP_ALWAYS(*inBuff);
1193
1194 const size_t numChars = decode_utf16(tmp, ch);
1195 if ( numChars == wxCONV_FAILED )
1196 return wxCONV_FAILED;
1197
1198 if ( numChars == 2 )
1199 inBuff++;
1200
1201 if ( ++outLen > dstLen )
1202 return wxCONV_FAILED;
1203
1204 *dst++ = ch;
1205 }
1206
1207
1208 return outLen;
1209}
1210
1211size_t
1212wxMBConvUTF16swap::FromWChar(char *dst, size_t dstLen,
1213 const wchar_t *src, size_t srcLen) const
1214{
1215 if ( srcLen == wxNO_LEN )
1216 srcLen = wxWcslen(src) + 1;
1217
1218 size_t outLen = 0;
1219 wxUint16 *outBuff = wx_reinterpret_cast(wxUint16 *, dst);
1220 for ( const wchar_t *srcEnd = src + srcLen; src < srcEnd; src++ )
1221 {
1222 wxUint16 cc[2];
1223 const size_t numChars = encode_utf16(*src, cc);
1224 if ( numChars == wxCONV_FAILED )
1225 return wxCONV_FAILED;
1226
1227 outLen += numChars * BYTES_PER_CHAR;
1228 if ( outBuff )
1229 {
1230 if ( outLen > dstLen )
1231 return wxCONV_FAILED;
1232
1233 *outBuff++ = wxUINT16_SWAP_ALWAYS(cc[0]);
1234 if ( numChars == 2 )
1235 {
1236 // second character of a surrogate
1237 *outBuff++ = wxUINT16_SWAP_ALWAYS(cc[1]);
1238 }
1239 }
1240 }
1241
1242 return outLen;
1243}
1244
1245#endif // WC_UTF16/!WC_UTF16
1246
1247
1248// ============================================================================
1249// UTF-32
1250// ============================================================================
1251
1252#ifdef WORDS_BIGENDIAN
1253 #define wxMBConvUTF32straight wxMBConvUTF32BE
1254 #define wxMBConvUTF32swap wxMBConvUTF32LE
1255#else
1256 #define wxMBConvUTF32swap wxMBConvUTF32BE
1257 #define wxMBConvUTF32straight wxMBConvUTF32LE
1258#endif
1259
1260
1261WXDLLIMPEXP_DATA_BASE(wxMBConvUTF32LE) wxConvUTF32LE;
1262WXDLLIMPEXP_DATA_BASE(wxMBConvUTF32BE) wxConvUTF32BE;
1263
1264/* static */
1265size_t wxMBConvUTF32Base::GetLength(const char *src, size_t srcLen)
1266{
1267 if ( srcLen == wxNO_LEN )
1268 {
1269 // count the number of bytes in input, including the trailing NULs
1270 const wxUint32 *inBuff = wx_reinterpret_cast(const wxUint32 *, src);
1271 for ( srcLen = 1; *inBuff++; srcLen++ )
1272 ;
1273
1274 srcLen *= BYTES_PER_CHAR;
1275 }
1276 else // we already have the length
1277 {
1278 // we can only convert an entire number of UTF-32 characters
1279 if ( srcLen % BYTES_PER_CHAR )
1280 return wxCONV_FAILED;
1281 }
1282
1283 return srcLen;
1284}
1285
1286// case when in-memory representation is UTF-16
1287#ifdef WC_UTF16
1288
1289// ----------------------------------------------------------------------------
1290// conversions without endianness change
1291// ----------------------------------------------------------------------------
1292
1293size_t
1294wxMBConvUTF32straight::ToWChar(wchar_t *dst, size_t dstLen,
1295 const char *src, size_t srcLen) const
1296{
1297 srcLen = GetLength(src, srcLen);
1298 if ( srcLen == wxNO_LEN )
1299 return wxCONV_FAILED;
1300
1301 const wxUint32 *inBuff = wx_reinterpret_cast(const wxUint32 *, src);
1302 const size_t inLen = srcLen / BYTES_PER_CHAR;
1303 size_t outLen = 0;
1304 for ( size_t n = 0; n < inLen; n++ )
1305 {
1306 wxUint16 cc[2];
1307 const size_t numChars = encode_utf16(*inBuff++, cc);
1308 if ( numChars == wxCONV_FAILED )
1309 return wxCONV_FAILED;
1310
1311 outLen += numChars;
1312 if ( dst )
1313 {
1314 if ( outLen > dstLen )
1315 return wxCONV_FAILED;
1316
1317 *dst++ = cc[0];
1318 if ( numChars == 2 )
1319 {
1320 // second character of a surrogate
1321 *dst++ = cc[1];
1322 }
1323 }
1324 }
1325
1326 return outLen;
1327}
1328
1329size_t
1330wxMBConvUTF32straight::FromWChar(char *dst, size_t dstLen,
1331 const wchar_t *src, size_t srcLen) const
1332{
1333 if ( srcLen == wxNO_LEN )
1334 srcLen = wxWcslen(src) + 1;
1335
1336 if ( !dst )
1337 {
1338 // optimization: return maximal space which could be needed for this
1339 // string instead of the exact amount which could be less if there are
1340 // any surrogates in the input
1341 //
1342 // we consider that surrogates are rare enough to make it worthwhile to
1343 // avoid running the loop below at the cost of slightly extra memory
1344 // consumption
1345 return srcLen * BYTES_PER_CHAR;
1346 }
1347
1348 wxUint32 *outBuff = wx_reinterpret_cast(wxUint32 *, dst);
1349 size_t outLen = 0;
1350 for ( const wchar_t * const srcEnd = src + srcLen; src < srcEnd; )
1351 {
1352 const wxUint32 ch = wxDecodeSurrogate(&src);
1353 if ( !src )
1354 return wxCONV_FAILED;
1355
1356 outLen += BYTES_PER_CHAR;
1357
1358 if ( outLen > dstLen )
1359 return wxCONV_FAILED;
1360
1361 *outBuff++ = ch;
1362 }
1363
1364 return outLen;
1365}
1366
1367// ----------------------------------------------------------------------------
1368// endian-reversing conversions
1369// ----------------------------------------------------------------------------
1370
1371size_t
1372wxMBConvUTF32swap::ToWChar(wchar_t *dst, size_t dstLen,
1373 const char *src, size_t srcLen) const
1374{
1375 srcLen = GetLength(src, srcLen);
1376 if ( srcLen == wxNO_LEN )
1377 return wxCONV_FAILED;
1378
1379 const wxUint32 *inBuff = wx_reinterpret_cast(const wxUint32 *, src);
1380 const size_t inLen = srcLen / BYTES_PER_CHAR;
1381 size_t outLen = 0;
1382 for ( size_t n = 0; n < inLen; n++, inBuff++ )
1383 {
1384 wxUint16 cc[2];
1385 const size_t numChars = encode_utf16(wxUINT32_SWAP_ALWAYS(*inBuff), cc);
1386 if ( numChars == wxCONV_FAILED )
1387 return wxCONV_FAILED;
1388
1389 outLen += numChars;
1390 if ( dst )
1391 {
1392 if ( outLen > dstLen )
1393 return wxCONV_FAILED;
1394
1395 *dst++ = cc[0];
1396 if ( numChars == 2 )
1397 {
1398 // second character of a surrogate
1399 *dst++ = cc[1];
1400 }
1401 }
1402 }
1403
1404 return outLen;
1405}
1406
1407size_t
1408wxMBConvUTF32swap::FromWChar(char *dst, size_t dstLen,
1409 const wchar_t *src, size_t srcLen) const
1410{
1411 if ( srcLen == wxNO_LEN )
1412 srcLen = wxWcslen(src) + 1;
1413
1414 if ( !dst )
1415 {
1416 // optimization: return maximal space which could be needed for this
1417 // string instead of the exact amount which could be less if there are
1418 // any surrogates in the input
1419 //
1420 // we consider that surrogates are rare enough to make it worthwhile to
1421 // avoid running the loop below at the cost of slightly extra memory
1422 // consumption
1423 return srcLen*BYTES_PER_CHAR;
1424 }
1425
1426 wxUint32 *outBuff = wx_reinterpret_cast(wxUint32 *, dst);
1427 size_t outLen = 0;
1428 for ( const wchar_t * const srcEnd = src + srcLen; src < srcEnd; )
1429 {
1430 const wxUint32 ch = wxDecodeSurrogate(&src);
1431 if ( !src )
1432 return wxCONV_FAILED;
1433
1434 outLen += BYTES_PER_CHAR;
1435
1436 if ( outLen > dstLen )
1437 return wxCONV_FAILED;
1438
1439 *outBuff++ = wxUINT32_SWAP_ALWAYS(ch);
1440 }
1441
1442 return outLen;
1443}
1444
1445#else // !WC_UTF16: wchar_t is UTF-32
1446
1447// ----------------------------------------------------------------------------
1448// conversions without endianness change
1449// ----------------------------------------------------------------------------
1450
1451size_t
1452wxMBConvUTF32straight::ToWChar(wchar_t *dst, size_t dstLen,
1453 const char *src, size_t srcLen) const
1454{
1455 // use memcpy() as it should be much faster than hand-written loop
1456 srcLen = GetLength(src, srcLen);
1457 if ( srcLen == wxNO_LEN )
1458 return wxCONV_FAILED;
1459
1460 const size_t inLen = srcLen/BYTES_PER_CHAR;
1461 if ( dst )
1462 {
1463 if ( dstLen < inLen )
1464 return wxCONV_FAILED;
1465
1466 memcpy(dst, src, srcLen);
1467 }
1468
1469 return inLen;
1470}
1471
1472size_t
1473wxMBConvUTF32straight::FromWChar(char *dst, size_t dstLen,
1474 const wchar_t *src, size_t srcLen) const
1475{
1476 if ( srcLen == wxNO_LEN )
1477 srcLen = wxWcslen(src) + 1;
1478
1479 srcLen *= BYTES_PER_CHAR;
1480
1481 if ( dst )
1482 {
1483 if ( dstLen < srcLen )
1484 return wxCONV_FAILED;
1485
1486 memcpy(dst, src, srcLen);
1487 }
1488
1489 return srcLen;
1490}
1491
1492// ----------------------------------------------------------------------------
1493// endian-reversing conversions
1494// ----------------------------------------------------------------------------
1495
1496size_t
1497wxMBConvUTF32swap::ToWChar(wchar_t *dst, size_t dstLen,
1498 const char *src, size_t srcLen) const
1499{
1500 srcLen = GetLength(src, srcLen);
1501 if ( srcLen == wxNO_LEN )
1502 return wxCONV_FAILED;
1503
1504 srcLen /= BYTES_PER_CHAR;
1505
1506 if ( dst )
1507 {
1508 if ( dstLen < srcLen )
1509 return wxCONV_FAILED;
1510
1511 const wxUint32 *inBuff = wx_reinterpret_cast(const wxUint32 *, src);
1512 for ( size_t n = 0; n < srcLen; n++, inBuff++ )
1513 {
1514 *dst++ = wxUINT32_SWAP_ALWAYS(*inBuff);
1515 }
1516 }
1517
1518 return srcLen;
1519}
1520
1521size_t
1522wxMBConvUTF32swap::FromWChar(char *dst, size_t dstLen,
1523 const wchar_t *src, size_t srcLen) const
1524{
1525 if ( srcLen == wxNO_LEN )
1526 srcLen = wxWcslen(src) + 1;
1527
1528 srcLen *= BYTES_PER_CHAR;
1529
1530 if ( dst )
1531 {
1532 if ( dstLen < srcLen )
1533 return wxCONV_FAILED;
1534
1535 wxUint32 *outBuff = wx_reinterpret_cast(wxUint32 *, dst);
1536 for ( size_t n = 0; n < srcLen; n += BYTES_PER_CHAR, src++ )
1537 {
1538 *outBuff++ = wxUINT32_SWAP_ALWAYS(*src);
1539 }
1540 }
1541
1542 return srcLen;
1543}
1544
1545#endif // WC_UTF16/!WC_UTF16
1546
1547
1548// ============================================================================
1549// The classes doing conversion using the iconv_xxx() functions
1550// ============================================================================
1551
1552#ifdef HAVE_ICONV
1553
1554// VS: glibc 2.1.3 is broken in that iconv() conversion to/from UCS4 fails with
1555// E2BIG if output buffer is _exactly_ as big as needed. Such case is
1556// (unless there's yet another bug in glibc) the only case when iconv()
1557// returns with (size_t)-1 (which means error) and says there are 0 bytes
1558// left in the input buffer -- when _real_ error occurs,
1559// bytes-left-in-input buffer is non-zero. Hence, this alternative test for
1560// iconv() failure.
1561// [This bug does not appear in glibc 2.2.]
1562#if defined(__GLIBC__) && __GLIBC__ == 2 && __GLIBC_MINOR__ <= 1
1563#define ICONV_FAILED(cres, bufLeft) ((cres == (size_t)-1) && \
1564 (errno != E2BIG || bufLeft != 0))
1565#else
1566#define ICONV_FAILED(cres, bufLeft) (cres == (size_t)-1)
1567#endif
1568
1569#define ICONV_CHAR_CAST(x) ((ICONV_CONST char **)(x))
1570
1571#define ICONV_T_INVALID ((iconv_t)-1)
1572
1573#if SIZEOF_WCHAR_T == 4
1574 #define WC_BSWAP wxUINT32_SWAP_ALWAYS
1575 #define WC_ENC wxFONTENCODING_UTF32
1576#elif SIZEOF_WCHAR_T == 2
1577 #define WC_BSWAP wxUINT16_SWAP_ALWAYS
1578 #define WC_ENC wxFONTENCODING_UTF16
1579#else // sizeof(wchar_t) != 2 nor 4
1580 // does this ever happen?
1581 #error "Unknown sizeof(wchar_t): please report this to wx-dev@lists.wxwindows.org"
1582#endif
1583
1584// ----------------------------------------------------------------------------
1585// wxMBConv_iconv: encapsulates an iconv character set
1586// ----------------------------------------------------------------------------
1587
1588class wxMBConv_iconv : public wxMBConv
1589{
1590public:
1591 wxMBConv_iconv(const char *name);
1592 virtual ~wxMBConv_iconv();
1593
1594 virtual size_t MB2WC(wchar_t *buf, const char *psz, size_t n) const;
1595 virtual size_t WC2MB(char *buf, const wchar_t *psz, size_t n) const;
1596
1597 // classify this encoding as explained in wxMBConv::GetMBNulLen() comment
1598 virtual size_t GetMBNulLen() const;
1599
1600#if wxUSE_UNICODE_UTF8
1601 virtual bool IsUTF8() const;
1602#endif
1603
1604 virtual wxMBConv *Clone() const
1605 {
1606 wxMBConv_iconv *p = new wxMBConv_iconv(m_name.ToAscii());
1607 p->m_minMBCharWidth = m_minMBCharWidth;
1608 return p;
1609 }
1610
1611 bool IsOk() const
1612 { return (m2w != ICONV_T_INVALID) && (w2m != ICONV_T_INVALID); }
1613
1614protected:
1615 // the iconv handlers used to translate from multibyte
1616 // to wide char and in the other direction
1617 iconv_t m2w,
1618 w2m;
1619
1620#if wxUSE_THREADS
1621 // guards access to m2w and w2m objects
1622 wxMutex m_iconvMutex;
1623#endif
1624
1625private:
1626 // the name (for iconv_open()) of a wide char charset -- if none is
1627 // available on this machine, it will remain NULL
1628 static wxString ms_wcCharsetName;
1629
1630 // true if the wide char encoding we use (i.e. ms_wcCharsetName) has
1631 // different endian-ness than the native one
1632 static bool ms_wcNeedsSwap;
1633
1634
1635 // name of the encoding handled by this conversion
1636 wxString m_name;
1637
1638 // cached result of GetMBNulLen(); set to 0 meaning "unknown"
1639 // initially
1640 size_t m_minMBCharWidth;
1641};
1642
1643// make the constructor available for unit testing
1644WXDLLIMPEXP_BASE wxMBConv* new_wxMBConv_iconv( const char* name )
1645{
1646 wxMBConv_iconv* result = new wxMBConv_iconv( name );
1647 if ( !result->IsOk() )
1648 {
1649 delete result;
1650 return 0;
1651 }
1652
1653 return result;
1654}
1655
1656wxString wxMBConv_iconv::ms_wcCharsetName;
1657bool wxMBConv_iconv::ms_wcNeedsSwap = false;
1658
1659wxMBConv_iconv::wxMBConv_iconv(const char *name)
1660 : m_name(name)
1661{
1662 m_minMBCharWidth = 0;
1663
1664 // check for charset that represents wchar_t:
1665 if ( ms_wcCharsetName.empty() )
1666 {
1667 wxLogTrace(TRACE_STRCONV, _T("Looking for wide char codeset:"));
1668
1669#if wxUSE_FONTMAP
1670 const wxChar **names = wxFontMapperBase::GetAllEncodingNames(WC_ENC);
1671#else // !wxUSE_FONTMAP
1672 static const wxChar *names_static[] =
1673 {
1674#if SIZEOF_WCHAR_T == 4
1675 _T("UCS-4"),
1676#elif SIZEOF_WCHAR_T = 2
1677 _T("UCS-2"),
1678#endif
1679 NULL
1680 };
1681 const wxChar **names = names_static;
1682#endif // wxUSE_FONTMAP/!wxUSE_FONTMAP
1683
1684 for ( ; *names && ms_wcCharsetName.empty(); ++names )
1685 {
1686 const wxString nameCS(*names);
1687
1688 // first try charset with explicit bytesex info (e.g. "UCS-4LE"):
1689 wxString nameXE(nameCS);
1690
1691#ifdef WORDS_BIGENDIAN
1692 nameXE += _T("BE");
1693#else // little endian
1694 nameXE += _T("LE");
1695#endif
1696
1697 wxLogTrace(TRACE_STRCONV, _T(" trying charset \"%s\""),
1698 nameXE.c_str());
1699
1700 m2w = iconv_open(nameXE.ToAscii(), name);
1701 if ( m2w == ICONV_T_INVALID )
1702 {
1703 // try charset w/o bytesex info (e.g. "UCS4")
1704 wxLogTrace(TRACE_STRCONV, _T(" trying charset \"%s\""),
1705 nameCS.c_str());
1706 m2w = iconv_open(nameCS.ToAscii(), name);
1707
1708 // and check for bytesex ourselves:
1709 if ( m2w != ICONV_T_INVALID )
1710 {
1711 char buf[2], *bufPtr;
1712 wchar_t wbuf[2], *wbufPtr;
1713 size_t insz, outsz;
1714 size_t res;
1715
1716 buf[0] = 'A';
1717 buf[1] = 0;
1718 wbuf[0] = 0;
1719 insz = 2;
1720 outsz = SIZEOF_WCHAR_T * 2;
1721 wbufPtr = wbuf;
1722 bufPtr = buf;
1723
1724 res = iconv(
1725 m2w, ICONV_CHAR_CAST(&bufPtr), &insz,
1726 (char**)&wbufPtr, &outsz);
1727
1728 if (ICONV_FAILED(res, insz))
1729 {
1730 wxLogLastError(wxT("iconv"));
1731 wxLogError(_("Conversion to charset '%s' doesn't work."),
1732 nameCS.c_str());
1733 }
1734 else // ok, can convert to this encoding, remember it
1735 {
1736 ms_wcCharsetName = nameCS;
1737 ms_wcNeedsSwap = wbuf[0] != (wchar_t)buf[0];
1738 }
1739 }
1740 }
1741 else // use charset not requiring byte swapping
1742 {
1743 ms_wcCharsetName = nameXE;
1744 }
1745 }
1746
1747 wxLogTrace(TRACE_STRCONV,
1748 wxT("iconv wchar_t charset is \"%s\"%s"),
1749 ms_wcCharsetName.empty() ? wxString("<none>")
1750 : ms_wcCharsetName,
1751 ms_wcNeedsSwap ? _T(" (needs swap)")
1752 : _T(""));
1753 }
1754 else // we already have ms_wcCharsetName
1755 {
1756 m2w = iconv_open(ms_wcCharsetName.ToAscii(), name);
1757 }
1758
1759 if ( ms_wcCharsetName.empty() )
1760 {
1761 w2m = ICONV_T_INVALID;
1762 }
1763 else
1764 {
1765 w2m = iconv_open(name, ms_wcCharsetName.ToAscii());
1766 if ( w2m == ICONV_T_INVALID )
1767 {
1768 wxLogTrace(TRACE_STRCONV,
1769 wxT("\"%s\" -> \"%s\" works but not the converse!?"),
1770 ms_wcCharsetName.c_str(), name);
1771 }
1772 }
1773}
1774
1775wxMBConv_iconv::~wxMBConv_iconv()
1776{
1777 if ( m2w != ICONV_T_INVALID )
1778 iconv_close(m2w);
1779 if ( w2m != ICONV_T_INVALID )
1780 iconv_close(w2m);
1781}
1782
1783size_t wxMBConv_iconv::MB2WC(wchar_t *buf, const char *psz, size_t n) const
1784{
1785 // find the string length: notice that must be done differently for
1786 // NUL-terminated strings and UTF-16/32 which are terminated with 2/4 NULs
1787 size_t inbuf;
1788 const size_t nulLen = GetMBNulLen();
1789 switch ( nulLen )
1790 {
1791 default:
1792 return wxCONV_FAILED;
1793
1794 case 1:
1795 inbuf = strlen(psz); // arguably more optimized than our version
1796 break;
1797
1798 case 2:
1799 case 4:
1800 // for UTF-16/32 not only we need to have 2/4 consecutive NULs but
1801 // they also have to start at character boundary and not span two
1802 // adjacent characters
1803 const char *p;
1804 for ( p = psz; NotAllNULs(p, nulLen); p += nulLen )
1805 ;
1806 inbuf = p - psz;
1807 break;
1808 }
1809
1810#if wxUSE_THREADS
1811 // NB: iconv() is MT-safe, but each thread must use its own iconv_t handle.
1812 // Unfortunately there are a couple of global wxCSConv objects such as
1813 // wxConvLocal that are used all over wx code, so we have to make sure
1814 // the handle is used by at most one thread at the time. Otherwise
1815 // only a few wx classes would be safe to use from non-main threads
1816 // as MB<->WC conversion would fail "randomly".
1817 wxMutexLocker lock(wxConstCast(this, wxMBConv_iconv)->m_iconvMutex);
1818#endif // wxUSE_THREADS
1819
1820 size_t outbuf = n * SIZEOF_WCHAR_T;
1821 size_t res, cres;
1822 // VS: Use these instead of psz, buf because iconv() modifies its arguments:
1823 wchar_t *bufPtr = buf;
1824 const char *pszPtr = psz;
1825
1826 if (buf)
1827 {
1828 // have destination buffer, convert there
1829 cres = iconv(m2w,
1830 ICONV_CHAR_CAST(&pszPtr), &inbuf,
1831 (char**)&bufPtr, &outbuf);
1832 res = n - (outbuf / SIZEOF_WCHAR_T);
1833
1834 if (ms_wcNeedsSwap)
1835 {
1836 // convert to native endianness
1837 for ( unsigned i = 0; i < res; i++ )
1838 buf[n] = WC_BSWAP(buf[i]);
1839 }
1840
1841 // NUL-terminate the string if there is any space left
1842 if (res < n)
1843 buf[res] = 0;
1844 }
1845 else
1846 {
1847 // no destination buffer... convert using temp buffer
1848 // to calculate destination buffer requirement
1849 wchar_t tbuf[8];
1850 res = 0;
1851
1852 do
1853 {
1854 bufPtr = tbuf;
1855 outbuf = 8 * SIZEOF_WCHAR_T;
1856
1857 cres = iconv(m2w,
1858 ICONV_CHAR_CAST(&pszPtr), &inbuf,
1859 (char**)&bufPtr, &outbuf );
1860
1861 res += 8 - (outbuf / SIZEOF_WCHAR_T);
1862 }
1863 while ((cres == (size_t)-1) && (errno == E2BIG));
1864 }
1865
1866 if (ICONV_FAILED(cres, inbuf))
1867 {
1868 //VS: it is ok if iconv fails, hence trace only
1869 wxLogTrace(TRACE_STRCONV, wxT("iconv failed: %s"), wxSysErrorMsg(wxSysErrorCode()));
1870 return wxCONV_FAILED;
1871 }
1872
1873 return res;
1874}
1875
1876size_t wxMBConv_iconv::WC2MB(char *buf, const wchar_t *psz, size_t n) const
1877{
1878#if wxUSE_THREADS
1879 // NB: explained in MB2WC
1880 wxMutexLocker lock(wxConstCast(this, wxMBConv_iconv)->m_iconvMutex);
1881#endif
1882
1883 size_t inlen = wxWcslen(psz);
1884 size_t inbuf = inlen * SIZEOF_WCHAR_T;
1885 size_t outbuf = n;
1886 size_t res, cres;
1887
1888 wchar_t *tmpbuf = 0;
1889
1890 if (ms_wcNeedsSwap)
1891 {
1892 // need to copy to temp buffer to switch endianness
1893 // (doing WC_BSWAP twice on the original buffer won't help, as it
1894 // could be in read-only memory, or be accessed in some other thread)
1895 tmpbuf = (wchar_t *)malloc(inbuf + SIZEOF_WCHAR_T);
1896 for ( size_t i = 0; i < inlen; i++ )
1897 tmpbuf[n] = WC_BSWAP(psz[i]);
1898
1899 tmpbuf[inlen] = L'\0';
1900 psz = tmpbuf;
1901 }
1902
1903 if (buf)
1904 {
1905 // have destination buffer, convert there
1906 cres = iconv( w2m, ICONV_CHAR_CAST(&psz), &inbuf, &buf, &outbuf );
1907
1908 res = n - outbuf;
1909
1910 // NB: iconv was given only wcslen(psz) characters on input, and so
1911 // it couldn't convert the trailing zero. Let's do it ourselves
1912 // if there's some room left for it in the output buffer.
1913 if (res < n)
1914 buf[0] = 0;
1915 }
1916 else
1917 {
1918 // no destination buffer: convert using temp buffer
1919 // to calculate destination buffer requirement
1920 char tbuf[16];
1921 res = 0;
1922 do
1923 {
1924 buf = tbuf;
1925 outbuf = 16;
1926
1927 cres = iconv( w2m, ICONV_CHAR_CAST(&psz), &inbuf, &buf, &outbuf );
1928
1929 res += 16 - outbuf;
1930 }
1931 while ((cres == (size_t)-1) && (errno == E2BIG));
1932 }
1933
1934 if (ms_wcNeedsSwap)
1935 {
1936 free(tmpbuf);
1937 }
1938
1939 if (ICONV_FAILED(cres, inbuf))
1940 {
1941 wxLogTrace(TRACE_STRCONV, wxT("iconv failed: %s"), wxSysErrorMsg(wxSysErrorCode()));
1942 return wxCONV_FAILED;
1943 }
1944
1945 return res;
1946}
1947
1948size_t wxMBConv_iconv::GetMBNulLen() const
1949{
1950 if ( m_minMBCharWidth == 0 )
1951 {
1952 wxMBConv_iconv * const self = wxConstCast(this, wxMBConv_iconv);
1953
1954#if wxUSE_THREADS
1955 // NB: explained in MB2WC
1956 wxMutexLocker lock(self->m_iconvMutex);
1957#endif
1958
1959 const wchar_t *wnul = L"";
1960 char buf[8]; // should be enough for NUL in any encoding
1961 size_t inLen = sizeof(wchar_t),
1962 outLen = WXSIZEOF(buf);
1963 char *inBuff = (char *)wnul;
1964 char *outBuff = buf;
1965 if ( iconv(w2m, ICONV_CHAR_CAST(&inBuff), &inLen, &outBuff, &outLen) == (size_t)-1 )
1966 {
1967 self->m_minMBCharWidth = (size_t)-1;
1968 }
1969 else // ok
1970 {
1971 self->m_minMBCharWidth = outBuff - buf;
1972 }
1973 }
1974
1975 return m_minMBCharWidth;
1976}
1977
1978#if wxUSE_UNICODE_UTF8
1979bool wxMBConv_iconv::IsUTF8() const
1980{
1981 return wxStricmp(m_name, "UTF-8") == 0 ||
1982 wxStricmp(m_name, "UTF8") == 0;
1983}
1984#endif
1985
1986#endif // HAVE_ICONV
1987
1988
1989// ============================================================================
1990// Win32 conversion classes
1991// ============================================================================
1992
1993#ifdef wxHAVE_WIN32_MB2WC
1994
1995// from utils.cpp
1996#if wxUSE_FONTMAP
1997extern WXDLLIMPEXP_BASE long wxCharsetToCodepage(const char *charset);
1998extern WXDLLIMPEXP_BASE long wxEncodingToCodepage(wxFontEncoding encoding);
1999#endif
2000
2001class wxMBConv_win32 : public wxMBConv
2002{
2003public:
2004 wxMBConv_win32()
2005 {
2006 m_CodePage = CP_ACP;
2007 m_minMBCharWidth = 0;
2008 }
2009
2010 wxMBConv_win32(const wxMBConv_win32& conv)
2011 : wxMBConv()
2012 {
2013 m_CodePage = conv.m_CodePage;
2014 m_minMBCharWidth = conv.m_minMBCharWidth;
2015 }
2016
2017#if wxUSE_FONTMAP
2018 wxMBConv_win32(const char* name)
2019 {
2020 m_CodePage = wxCharsetToCodepage(name);
2021 m_minMBCharWidth = 0;
2022 }
2023
2024 wxMBConv_win32(wxFontEncoding encoding)
2025 {
2026 m_CodePage = wxEncodingToCodepage(encoding);
2027 m_minMBCharWidth = 0;
2028 }
2029#endif // wxUSE_FONTMAP
2030
2031 virtual size_t MB2WC(wchar_t *buf, const char *psz, size_t n) const
2032 {
2033 // note that we have to use MB_ERR_INVALID_CHARS flag as it without it
2034 // the behaviour is not compatible with the Unix version (using iconv)
2035 // and break the library itself, e.g. wxTextInputStream::NextChar()
2036 // wouldn't work if reading an incomplete MB char didn't result in an
2037 // error
2038 //
2039 // Moreover, MB_ERR_INVALID_CHARS is only supported on Win 2K SP4 or
2040 // Win XP or newer and it is not supported for UTF-[78] so we always
2041 // use our own conversions in this case. See
2042 // http://blogs.msdn.com/michkap/archive/2005/04/19/409566.aspx
2043 // http://msdn.microsoft.com/library/en-us/intl/unicode_17si.asp
2044 if ( m_CodePage == CP_UTF8 )
2045 {
2046 return wxMBConvUTF8().MB2WC(buf, psz, n);
2047 }
2048
2049 if ( m_CodePage == CP_UTF7 )
2050 {
2051 return wxMBConvUTF7().MB2WC(buf, psz, n);
2052 }
2053
2054 int flags = 0;
2055 if ( (m_CodePage < 50000 && m_CodePage != CP_SYMBOL) &&
2056 IsAtLeastWin2kSP4() )
2057 {
2058 flags = MB_ERR_INVALID_CHARS;
2059 }
2060
2061 const size_t len = ::MultiByteToWideChar
2062 (
2063 m_CodePage, // code page
2064 flags, // flags: fall on error
2065 psz, // input string
2066 -1, // its length (NUL-terminated)
2067 buf, // output string
2068 buf ? n : 0 // size of output buffer
2069 );
2070 if ( !len )
2071 {
2072 // function totally failed
2073 return wxCONV_FAILED;
2074 }
2075
2076 // if we were really converting and didn't use MB_ERR_INVALID_CHARS,
2077 // check if we succeeded, by doing a double trip:
2078 if ( !flags && buf )
2079 {
2080 const size_t mbLen = strlen(psz);
2081 wxCharBuffer mbBuf(mbLen);
2082 if ( ::WideCharToMultiByte
2083 (
2084 m_CodePage,
2085 0,
2086 buf,
2087 -1,
2088 mbBuf.data(),
2089 mbLen + 1, // size in bytes, not length
2090 NULL,
2091 NULL
2092 ) == 0 ||
2093 strcmp(mbBuf, psz) != 0 )
2094 {
2095 // we didn't obtain the same thing we started from, hence
2096 // the conversion was lossy and we consider that it failed
2097 return wxCONV_FAILED;
2098 }
2099 }
2100
2101 // note that it returns count of written chars for buf != NULL and size
2102 // of the needed buffer for buf == NULL so in either case the length of
2103 // the string (which never includes the terminating NUL) is one less
2104 return len - 1;
2105 }
2106
2107 virtual size_t WC2MB(char *buf, const wchar_t *pwz, size_t n) const
2108 {
2109 /*
2110 we have a problem here: by default, WideCharToMultiByte() may
2111 replace characters unrepresentable in the target code page with bad
2112 quality approximations such as turning "1/2" symbol (U+00BD) into
2113 "1" for the code pages which don't have it and we, obviously, want
2114 to avoid this at any price
2115
2116 the trouble is that this function does it _silently_, i.e. it won't
2117 even tell us whether it did or not... Win98/2000 and higher provide
2118 WC_NO_BEST_FIT_CHARS but it doesn't work for the older systems and
2119 we have to resort to a round trip, i.e. check that converting back
2120 results in the same string -- this is, of course, expensive but
2121 otherwise we simply can't be sure to not garble the data.
2122 */
2123
2124 // determine if we can rely on WC_NO_BEST_FIT_CHARS: according to MSDN
2125 // it doesn't work with CJK encodings (which we test for rather roughly
2126 // here...) nor with UTF-7/8 nor, of course, with Windows versions not
2127 // supporting it
2128 BOOL usedDef wxDUMMY_INITIALIZE(false);
2129 BOOL *pUsedDef;
2130 int flags;
2131 if ( CanUseNoBestFit() && m_CodePage < 50000 )
2132 {
2133 // it's our lucky day
2134 flags = WC_NO_BEST_FIT_CHARS;
2135 pUsedDef = &usedDef;
2136 }
2137 else // old system or unsupported encoding
2138 {
2139 flags = 0;
2140 pUsedDef = NULL;
2141 }
2142
2143 const size_t len = ::WideCharToMultiByte
2144 (
2145 m_CodePage, // code page
2146 flags, // either none or no best fit
2147 pwz, // input string
2148 -1, // it is (wide) NUL-terminated
2149 buf, // output buffer
2150 buf ? n : 0, // and its size
2151 NULL, // default "replacement" char
2152 pUsedDef // [out] was it used?
2153 );
2154
2155 if ( !len )
2156 {
2157 // function totally failed
2158 return wxCONV_FAILED;
2159 }
2160
2161 // if we were really converting, check if we succeeded
2162 if ( buf )
2163 {
2164 if ( flags )
2165 {
2166 // check if the conversion failed, i.e. if any replacements
2167 // were done
2168 if ( usedDef )
2169 return wxCONV_FAILED;
2170 }
2171 else // we must resort to double tripping...
2172 {
2173 wxWCharBuffer wcBuf(n);
2174 if ( MB2WC(wcBuf.data(), buf, n) == wxCONV_FAILED ||
2175 wcscmp(wcBuf, pwz) != 0 )
2176 {
2177 // we didn't obtain the same thing we started from, hence
2178 // the conversion was lossy and we consider that it failed
2179 return wxCONV_FAILED;
2180 }
2181 }
2182 }
2183
2184 // see the comment above for the reason of "len - 1"
2185 return len - 1;
2186 }
2187
2188 virtual size_t GetMBNulLen() const
2189 {
2190 if ( m_minMBCharWidth == 0 )
2191 {
2192 int len = ::WideCharToMultiByte
2193 (
2194 m_CodePage, // code page
2195 0, // no flags
2196 L"", // input string
2197 1, // translate just the NUL
2198 NULL, // output buffer
2199 0, // and its size
2200 NULL, // no replacement char
2201 NULL // [out] don't care if it was used
2202 );
2203
2204 wxMBConv_win32 * const self = wxConstCast(this, wxMBConv_win32);
2205 switch ( len )
2206 {
2207 default:
2208 wxLogDebug(_T("Unexpected NUL length %d"), len);
2209 self->m_minMBCharWidth = (size_t)-1;
2210 break;
2211
2212 case 0:
2213 self->m_minMBCharWidth = (size_t)-1;
2214 break;
2215
2216 case 1:
2217 case 2:
2218 case 4:
2219 self->m_minMBCharWidth = len;
2220 break;
2221 }
2222 }
2223
2224 return m_minMBCharWidth;
2225 }
2226
2227 virtual wxMBConv *Clone() const { return new wxMBConv_win32(*this); }
2228
2229 bool IsOk() const { return m_CodePage != -1; }
2230
2231private:
2232 static bool CanUseNoBestFit()
2233 {
2234 static int s_isWin98Or2k = -1;
2235
2236 if ( s_isWin98Or2k == -1 )
2237 {
2238 int verMaj, verMin;
2239 switch ( wxGetOsVersion(&verMaj, &verMin) )
2240 {
2241 case wxOS_WINDOWS_9X:
2242 s_isWin98Or2k = verMaj >= 4 && verMin >= 10;
2243 break;
2244
2245 case wxOS_WINDOWS_NT:
2246 s_isWin98Or2k = verMaj >= 5;
2247 break;
2248
2249 default:
2250 // unknown: be conservative by default
2251 s_isWin98Or2k = 0;
2252 break;
2253 }
2254
2255 wxASSERT_MSG( s_isWin98Or2k != -1, _T("should be set above") );
2256 }
2257
2258 return s_isWin98Or2k == 1;
2259 }
2260
2261 static bool IsAtLeastWin2kSP4()
2262 {
2263#ifdef __WXWINCE__
2264 return false;
2265#else
2266 static int s_isAtLeastWin2kSP4 = -1;
2267
2268 if ( s_isAtLeastWin2kSP4 == -1 )
2269 {
2270 OSVERSIONINFOEX ver;
2271
2272 memset(&ver, 0, sizeof(ver));
2273 ver.dwOSVersionInfoSize = sizeof(ver);
2274 GetVersionEx((OSVERSIONINFO*)&ver);
2275
2276 s_isAtLeastWin2kSP4 =
2277 ((ver.dwMajorVersion > 5) || // Vista+
2278 (ver.dwMajorVersion == 5 && ver.dwMinorVersion > 0) || // XP/2003
2279 (ver.dwMajorVersion == 5 && ver.dwMinorVersion == 0 &&
2280 ver.wServicePackMajor >= 4)) // 2000 SP4+
2281 ? 1 : 0;
2282 }
2283
2284 return s_isAtLeastWin2kSP4 == 1;
2285#endif
2286 }
2287
2288
2289 // the code page we're working with
2290 long m_CodePage;
2291
2292 // cached result of GetMBNulLen(), set to 0 initially meaning
2293 // "unknown"
2294 size_t m_minMBCharWidth;
2295};
2296
2297#endif // wxHAVE_WIN32_MB2WC
2298
2299// ============================================================================
2300// Cocoa conversion classes
2301// ============================================================================
2302
2303#if defined(__WXCOCOA__)
2304
2305// RN: There is no UTF-32 support in either Core Foundation or Cocoa.
2306// Strangely enough, internally Core Foundation uses
2307// UTF-32 internally quite a bit - its just not public (yet).
2308
2309#include <CoreFoundation/CFString.h>
2310#include <CoreFoundation/CFStringEncodingExt.h>
2311
2312CFStringEncoding wxCFStringEncFromFontEnc(wxFontEncoding encoding)
2313{
2314 CFStringEncoding enc = kCFStringEncodingInvalidId ;
2315
2316 switch (encoding)
2317 {
2318 case wxFONTENCODING_DEFAULT :
2319 enc = CFStringGetSystemEncoding();
2320 break ;
2321
2322 case wxFONTENCODING_ISO8859_1 :
2323 enc = kCFStringEncodingISOLatin1 ;
2324 break ;
2325 case wxFONTENCODING_ISO8859_2 :
2326 enc = kCFStringEncodingISOLatin2;
2327 break ;
2328 case wxFONTENCODING_ISO8859_3 :
2329 enc = kCFStringEncodingISOLatin3 ;
2330 break ;
2331 case wxFONTENCODING_ISO8859_4 :
2332 enc = kCFStringEncodingISOLatin4;
2333 break ;
2334 case wxFONTENCODING_ISO8859_5 :
2335 enc = kCFStringEncodingISOLatinCyrillic;
2336 break ;
2337 case wxFONTENCODING_ISO8859_6 :
2338 enc = kCFStringEncodingISOLatinArabic;
2339 break ;
2340 case wxFONTENCODING_ISO8859_7 :
2341 enc = kCFStringEncodingISOLatinGreek;
2342 break ;
2343 case wxFONTENCODING_ISO8859_8 :
2344 enc = kCFStringEncodingISOLatinHebrew;
2345 break ;
2346 case wxFONTENCODING_ISO8859_9 :
2347 enc = kCFStringEncodingISOLatin5;
2348 break ;
2349 case wxFONTENCODING_ISO8859_10 :
2350 enc = kCFStringEncodingISOLatin6;
2351 break ;
2352 case wxFONTENCODING_ISO8859_11 :
2353 enc = kCFStringEncodingISOLatinThai;
2354 break ;
2355 case wxFONTENCODING_ISO8859_13 :
2356 enc = kCFStringEncodingISOLatin7;
2357 break ;
2358 case wxFONTENCODING_ISO8859_14 :
2359 enc = kCFStringEncodingISOLatin8;
2360 break ;
2361 case wxFONTENCODING_ISO8859_15 :
2362 enc = kCFStringEncodingISOLatin9;
2363 break ;
2364
2365 case wxFONTENCODING_KOI8 :
2366 enc = kCFStringEncodingKOI8_R;
2367 break ;
2368 case wxFONTENCODING_ALTERNATIVE : // MS-DOS CP866
2369 enc = kCFStringEncodingDOSRussian;
2370 break ;
2371
2372// case wxFONTENCODING_BULGARIAN :
2373// enc = ;
2374// break ;
2375
2376 case wxFONTENCODING_CP437 :
2377 enc = kCFStringEncodingDOSLatinUS ;
2378 break ;
2379 case wxFONTENCODING_CP850 :
2380 enc = kCFStringEncodingDOSLatin1;
2381 break ;
2382 case wxFONTENCODING_CP852 :
2383 enc = kCFStringEncodingDOSLatin2;
2384 break ;
2385 case wxFONTENCODING_CP855 :
2386 enc = kCFStringEncodingDOSCyrillic;
2387 break ;
2388 case wxFONTENCODING_CP866 :
2389 enc = kCFStringEncodingDOSRussian ;
2390 break ;
2391 case wxFONTENCODING_CP874 :
2392 enc = kCFStringEncodingDOSThai;
2393 break ;
2394 case wxFONTENCODING_CP932 :
2395 enc = kCFStringEncodingDOSJapanese;
2396 break ;
2397 case wxFONTENCODING_CP936 :
2398 enc = kCFStringEncodingDOSChineseSimplif ;
2399 break ;
2400 case wxFONTENCODING_CP949 :
2401 enc = kCFStringEncodingDOSKorean;
2402 break ;
2403 case wxFONTENCODING_CP950 :
2404 enc = kCFStringEncodingDOSChineseTrad;
2405 break ;
2406 case wxFONTENCODING_CP1250 :
2407 enc = kCFStringEncodingWindowsLatin2;
2408 break ;
2409 case wxFONTENCODING_CP1251 :
2410 enc = kCFStringEncodingWindowsCyrillic ;
2411 break ;
2412 case wxFONTENCODING_CP1252 :
2413 enc = kCFStringEncodingWindowsLatin1 ;
2414 break ;
2415 case wxFONTENCODING_CP1253 :
2416 enc = kCFStringEncodingWindowsGreek;
2417 break ;
2418 case wxFONTENCODING_CP1254 :
2419 enc = kCFStringEncodingWindowsLatin5;
2420 break ;
2421 case wxFONTENCODING_CP1255 :
2422 enc = kCFStringEncodingWindowsHebrew ;
2423 break ;
2424 case wxFONTENCODING_CP1256 :
2425 enc = kCFStringEncodingWindowsArabic ;
2426 break ;
2427 case wxFONTENCODING_CP1257 :
2428 enc = kCFStringEncodingWindowsBalticRim;
2429 break ;
2430// This only really encodes to UTF7 (if that) evidently
2431// case wxFONTENCODING_UTF7 :
2432// enc = kCFStringEncodingNonLossyASCII ;
2433// break ;
2434 case wxFONTENCODING_UTF8 :
2435 enc = kCFStringEncodingUTF8 ;
2436 break ;
2437 case wxFONTENCODING_EUC_JP :
2438 enc = kCFStringEncodingEUC_JP;
2439 break ;
2440 case wxFONTENCODING_UTF16 :
2441 enc = kCFStringEncodingUnicode ;
2442 break ;
2443 case wxFONTENCODING_MACROMAN :
2444 enc = kCFStringEncodingMacRoman ;
2445 break ;
2446 case wxFONTENCODING_MACJAPANESE :
2447 enc = kCFStringEncodingMacJapanese ;
2448 break ;
2449 case wxFONTENCODING_MACCHINESETRAD :
2450 enc = kCFStringEncodingMacChineseTrad ;
2451 break ;
2452 case wxFONTENCODING_MACKOREAN :
2453 enc = kCFStringEncodingMacKorean ;
2454 break ;
2455 case wxFONTENCODING_MACARABIC :
2456 enc = kCFStringEncodingMacArabic ;
2457 break ;
2458 case wxFONTENCODING_MACHEBREW :
2459 enc = kCFStringEncodingMacHebrew ;
2460 break ;
2461 case wxFONTENCODING_MACGREEK :
2462 enc = kCFStringEncodingMacGreek ;
2463 break ;
2464 case wxFONTENCODING_MACCYRILLIC :
2465 enc = kCFStringEncodingMacCyrillic ;
2466 break ;
2467 case wxFONTENCODING_MACDEVANAGARI :
2468 enc = kCFStringEncodingMacDevanagari ;
2469 break ;
2470 case wxFONTENCODING_MACGURMUKHI :
2471 enc = kCFStringEncodingMacGurmukhi ;
2472 break ;
2473 case wxFONTENCODING_MACGUJARATI :
2474 enc = kCFStringEncodingMacGujarati ;
2475 break ;
2476 case wxFONTENCODING_MACORIYA :
2477 enc = kCFStringEncodingMacOriya ;
2478 break ;
2479 case wxFONTENCODING_MACBENGALI :
2480 enc = kCFStringEncodingMacBengali ;
2481 break ;
2482 case wxFONTENCODING_MACTAMIL :
2483 enc = kCFStringEncodingMacTamil ;
2484 break ;
2485 case wxFONTENCODING_MACTELUGU :
2486 enc = kCFStringEncodingMacTelugu ;
2487 break ;
2488 case wxFONTENCODING_MACKANNADA :
2489 enc = kCFStringEncodingMacKannada ;
2490 break ;
2491 case wxFONTENCODING_MACMALAJALAM :
2492 enc = kCFStringEncodingMacMalayalam ;
2493 break ;
2494 case wxFONTENCODING_MACSINHALESE :
2495 enc = kCFStringEncodingMacSinhalese ;
2496 break ;
2497 case wxFONTENCODING_MACBURMESE :
2498 enc = kCFStringEncodingMacBurmese ;
2499 break ;
2500 case wxFONTENCODING_MACKHMER :
2501 enc = kCFStringEncodingMacKhmer ;
2502 break ;
2503 case wxFONTENCODING_MACTHAI :
2504 enc = kCFStringEncodingMacThai ;
2505 break ;
2506 case wxFONTENCODING_MACLAOTIAN :
2507 enc = kCFStringEncodingMacLaotian ;
2508 break ;
2509 case wxFONTENCODING_MACGEORGIAN :
2510 enc = kCFStringEncodingMacGeorgian ;
2511 break ;
2512 case wxFONTENCODING_MACARMENIAN :
2513 enc = kCFStringEncodingMacArmenian ;
2514 break ;
2515 case wxFONTENCODING_MACCHINESESIMP :
2516 enc = kCFStringEncodingMacChineseSimp ;
2517 break ;
2518 case wxFONTENCODING_MACTIBETAN :
2519 enc = kCFStringEncodingMacTibetan ;
2520 break ;
2521 case wxFONTENCODING_MACMONGOLIAN :
2522 enc = kCFStringEncodingMacMongolian ;
2523 break ;
2524 case wxFONTENCODING_MACETHIOPIC :
2525 enc = kCFStringEncodingMacEthiopic ;
2526 break ;
2527 case wxFONTENCODING_MACCENTRALEUR :
2528 enc = kCFStringEncodingMacCentralEurRoman ;
2529 break ;
2530 case wxFONTENCODING_MACVIATNAMESE :
2531 enc = kCFStringEncodingMacVietnamese ;
2532 break ;
2533 case wxFONTENCODING_MACARABICEXT :
2534 enc = kCFStringEncodingMacExtArabic ;
2535 break ;
2536 case wxFONTENCODING_MACSYMBOL :
2537 enc = kCFStringEncodingMacSymbol ;
2538 break ;
2539 case wxFONTENCODING_MACDINGBATS :
2540 enc = kCFStringEncodingMacDingbats ;
2541 break ;
2542 case wxFONTENCODING_MACTURKISH :
2543 enc = kCFStringEncodingMacTurkish ;
2544 break ;
2545 case wxFONTENCODING_MACCROATIAN :
2546 enc = kCFStringEncodingMacCroatian ;
2547 break ;
2548 case wxFONTENCODING_MACICELANDIC :
2549 enc = kCFStringEncodingMacIcelandic ;
2550 break ;
2551 case wxFONTENCODING_MACROMANIAN :
2552 enc = kCFStringEncodingMacRomanian ;
2553 break ;
2554 case wxFONTENCODING_MACCELTIC :
2555 enc = kCFStringEncodingMacCeltic ;
2556 break ;
2557 case wxFONTENCODING_MACGAELIC :
2558 enc = kCFStringEncodingMacGaelic ;
2559 break ;
2560// case wxFONTENCODING_MACKEYBOARD :
2561// enc = kCFStringEncodingMacKeyboardGlyphs ;
2562// break ;
2563
2564 default :
2565 // because gcc is picky
2566 break ;
2567 }
2568
2569 return enc ;
2570}
2571
2572class wxMBConv_cocoa : public wxMBConv
2573{
2574public:
2575 wxMBConv_cocoa()
2576 {
2577 Init(CFStringGetSystemEncoding()) ;
2578 }
2579
2580 wxMBConv_cocoa(const wxMBConv_cocoa& conv)
2581 {
2582 m_encoding = conv.m_encoding;
2583 }
2584
2585#if wxUSE_FONTMAP
2586 wxMBConv_cocoa(const wxChar* name)
2587 {
2588 Init( wxCFStringEncFromFontEnc(wxFontMapperBase::Get()->CharsetToEncoding(name, false) ) ) ;
2589 }
2590#endif
2591
2592 wxMBConv_cocoa(wxFontEncoding encoding)
2593 {
2594 Init( wxCFStringEncFromFontEnc(encoding) );
2595 }
2596
2597 virtual ~wxMBConv_cocoa()
2598 {
2599 }
2600
2601 void Init( CFStringEncoding encoding)
2602 {
2603 m_encoding = encoding ;
2604 }
2605
2606 size_t MB2WC(wchar_t * szOut, const char * szUnConv, size_t nOutSize) const
2607 {
2608 wxASSERT(szUnConv);
2609
2610 CFStringRef theString = CFStringCreateWithBytes (
2611 NULL, //the allocator
2612 (const UInt8*)szUnConv,
2613 strlen(szUnConv),
2614 m_encoding,
2615 false //no BOM/external representation
2616 );
2617
2618 wxASSERT(theString);
2619
2620 size_t nOutLength = CFStringGetLength(theString);
2621
2622 if (szOut == NULL)
2623 {
2624 CFRelease(theString);
2625 return nOutLength;
2626 }
2627
2628 CFRange theRange = { 0, nOutSize };
2629
2630#if SIZEOF_WCHAR_T == 4
2631 UniChar* szUniCharBuffer = new UniChar[nOutSize];
2632#endif
2633
2634 CFStringGetCharacters(theString, theRange, szUniCharBuffer);
2635
2636 CFRelease(theString);
2637
2638 szUniCharBuffer[nOutLength] = '\0';
2639
2640#if SIZEOF_WCHAR_T == 4
2641 wxMBConvUTF16 converter;
2642 converter.MB2WC( szOut, (const char*)szUniCharBuffer, nOutSize );
2643 delete [] szUniCharBuffer;
2644#endif
2645
2646 return nOutLength;
2647 }
2648
2649 size_t WC2MB(char *szOut, const wchar_t *szUnConv, size_t nOutSize) const
2650 {
2651 wxASSERT(szUnConv);
2652
2653 size_t nRealOutSize;
2654 size_t nBufSize = wxWcslen(szUnConv);
2655 UniChar* szUniBuffer = (UniChar*) szUnConv;
2656
2657#if SIZEOF_WCHAR_T == 4
2658 wxMBConvUTF16 converter ;
2659 nBufSize = converter.WC2MB( NULL, szUnConv, 0 );
2660 szUniBuffer = new UniChar[ (nBufSize / sizeof(UniChar)) + 1];
2661 converter.WC2MB( (char*) szUniBuffer, szUnConv, nBufSize + sizeof(UniChar));
2662 nBufSize /= sizeof(UniChar);
2663#endif
2664
2665 CFStringRef theString = CFStringCreateWithCharactersNoCopy(
2666 NULL, //allocator
2667 szUniBuffer,
2668 nBufSize,
2669 kCFAllocatorNull //deallocator - we want to deallocate it ourselves
2670 );
2671
2672 wxASSERT(theString);
2673
2674 //Note that CER puts a BOM when converting to unicode
2675 //so we check and use getchars instead in that case
2676 if (m_encoding == kCFStringEncodingUnicode)
2677 {
2678 if (szOut != NULL)
2679 CFStringGetCharacters(theString, CFRangeMake(0, nOutSize - 1), (UniChar*) szOut);
2680
2681 nRealOutSize = CFStringGetLength(theString) + 1;
2682 }
2683 else
2684 {
2685 CFStringGetBytes(
2686 theString,
2687 CFRangeMake(0, CFStringGetLength(theString)),
2688 m_encoding,
2689 0, //what to put in characters that can't be converted -
2690 //0 tells CFString to return NULL if it meets such a character
2691 false, //not an external representation
2692 (UInt8*) szOut,
2693 nOutSize,
2694 (CFIndex*) &nRealOutSize
2695 );
2696 }
2697
2698 CFRelease(theString);
2699
2700#if SIZEOF_WCHAR_T == 4
2701 delete[] szUniBuffer;
2702#endif
2703
2704 return nRealOutSize - 1;
2705 }
2706
2707 virtual wxMBConv *Clone() const { return new wxMBConv_cocoa(*this); }
2708
2709 bool IsOk() const
2710 {
2711 return m_encoding != kCFStringEncodingInvalidId &&
2712 CFStringIsEncodingAvailable(m_encoding);
2713 }
2714
2715private:
2716 CFStringEncoding m_encoding ;
2717};
2718
2719#endif // defined(__WXCOCOA__)
2720
2721// ============================================================================
2722// Mac conversion classes
2723// ============================================================================
2724
2725#if defined(__WXMAC__) && defined(TARGET_CARBON)
2726
2727class wxMBConv_mac : public wxMBConv
2728{
2729public:
2730 wxMBConv_mac()
2731 {
2732 Init(CFStringGetSystemEncoding()) ;
2733 }
2734
2735 wxMBConv_mac(const wxMBConv_mac& conv)
2736 {
2737 Init(conv.m_char_encoding);
2738 }
2739
2740#if wxUSE_FONTMAP
2741 wxMBConv_mac(const char* name)
2742 {
2743 Init( wxMacGetSystemEncFromFontEnc( wxFontMapperBase::Get()->CharsetToEncoding(name, false) ) );
2744 }
2745#endif
2746
2747 wxMBConv_mac(wxFontEncoding encoding)
2748 {
2749 Init( wxMacGetSystemEncFromFontEnc(encoding) );
2750 }
2751
2752 virtual ~wxMBConv_mac()
2753 {
2754 OSStatus status = noErr ;
2755 if (m_MB2WC_converter)
2756 status = TECDisposeConverter(m_MB2WC_converter);
2757 if (m_WC2MB_converter)
2758 status = TECDisposeConverter(m_WC2MB_converter);
2759 }
2760
2761 void Init( TextEncodingBase encoding,TextEncodingVariant encodingVariant = kTextEncodingDefaultVariant ,
2762 TextEncodingFormat encodingFormat = kTextEncodingDefaultFormat)
2763 {
2764 m_MB2WC_converter = NULL ;
2765 m_WC2MB_converter = NULL ;
2766 m_char_encoding = CreateTextEncoding(encoding, encodingVariant, encodingFormat) ;
2767 m_unicode_encoding = CreateTextEncoding(kTextEncodingUnicodeDefault, 0, kUnicode16BitFormat) ;
2768 }
2769
2770 virtual void CreateIfNeeded() const
2771 {
2772 if ( m_MB2WC_converter == NULL && m_WC2MB_converter == NULL )
2773 {
2774 OSStatus status = noErr ;
2775 status = TECCreateConverter(&m_MB2WC_converter,
2776 m_char_encoding,
2777 m_unicode_encoding);
2778 wxASSERT_MSG( status == noErr , _("Unable to create TextEncodingConverter")) ;
2779 status = TECCreateConverter(&m_WC2MB_converter,
2780 m_unicode_encoding,
2781 m_char_encoding);
2782 wxASSERT_MSG( status == noErr , _("Unable to create TextEncodingConverter")) ;
2783 }
2784 }
2785
2786 size_t MB2WC(wchar_t *buf, const char *psz, size_t n) const
2787 {
2788 CreateIfNeeded() ;
2789 OSStatus status = noErr ;
2790 ByteCount byteOutLen ;
2791 ByteCount byteInLen = strlen(psz) + 1;
2792 wchar_t *tbuf = NULL ;
2793 UniChar* ubuf = NULL ;
2794 size_t res = 0 ;
2795
2796 if (buf == NULL)
2797 {
2798 // Apple specs say at least 32
2799 n = wxMax( 32, byteInLen ) ;
2800 tbuf = (wchar_t*) malloc( n * SIZEOF_WCHAR_T ) ;
2801 }
2802
2803 ByteCount byteBufferLen = n * sizeof( UniChar ) ;
2804
2805#if SIZEOF_WCHAR_T == 4
2806 ubuf = (UniChar*) malloc( byteBufferLen + 2 ) ;
2807#else
2808 ubuf = (UniChar*) (buf ? buf : tbuf) ;
2809#endif
2810
2811 status = TECConvertText(
2812 m_MB2WC_converter, (ConstTextPtr) psz, byteInLen, &byteInLen,
2813 (TextPtr) ubuf, byteBufferLen, &byteOutLen);
2814
2815#if SIZEOF_WCHAR_T == 4
2816 // we have to terminate here, because n might be larger for the trailing zero, and if UniChar
2817 // is not properly terminated we get random characters at the end
2818 ubuf[byteOutLen / sizeof( UniChar ) ] = 0 ;
2819 wxMBConvUTF16 converter ;
2820 res = converter.MB2WC( (buf ? buf : tbuf), (const char*)ubuf, n ) ;
2821 free( ubuf ) ;
2822#else
2823 res = byteOutLen / sizeof( UniChar ) ;
2824#endif
2825
2826 if ( buf == NULL )
2827 free(tbuf) ;
2828
2829 if ( buf && res < n)
2830 buf[res] = 0;
2831
2832 return res ;
2833 }
2834
2835 size_t WC2MB(char *buf, const wchar_t *psz, size_t n) const
2836 {
2837 CreateIfNeeded() ;
2838 OSStatus status = noErr ;
2839 ByteCount byteOutLen ;
2840 ByteCount byteInLen = wxWcslen(psz) * SIZEOF_WCHAR_T ;
2841
2842 char *tbuf = NULL ;
2843
2844 if (buf == NULL)
2845 {
2846 // Apple specs say at least 32
2847 n = wxMax( 32, ((byteInLen / SIZEOF_WCHAR_T) * 8) + SIZEOF_WCHAR_T );
2848 tbuf = (char*) malloc( n ) ;
2849 }
2850
2851 ByteCount byteBufferLen = n ;
2852 UniChar* ubuf = NULL ;
2853
2854#if SIZEOF_WCHAR_T == 4
2855 wxMBConvUTF16 converter ;
2856 size_t unicharlen = converter.WC2MB( NULL, psz, 0 ) ;
2857 byteInLen = unicharlen ;
2858 ubuf = (UniChar*) malloc( byteInLen + 2 ) ;
2859 converter.WC2MB( (char*) ubuf, psz, unicharlen + 2 ) ;
2860#else
2861 ubuf = (UniChar*) psz ;
2862#endif
2863
2864 status = TECConvertText(
2865 m_WC2MB_converter, (ConstTextPtr) ubuf, byteInLen, &byteInLen,
2866 (TextPtr) (buf ? buf : tbuf), byteBufferLen, &byteOutLen);
2867
2868#if SIZEOF_WCHAR_T == 4
2869 free( ubuf ) ;
2870#endif
2871
2872 if ( buf == NULL )
2873 free(tbuf) ;
2874
2875 size_t res = byteOutLen ;
2876 if ( buf && res < n)
2877 {
2878 buf[res] = 0;
2879
2880 //we need to double-trip to verify it didn't insert any ? in place
2881 //of bogus characters
2882 wxWCharBuffer wcBuf(n);
2883 size_t pszlen = wxWcslen(psz);
2884 if ( MB2WC(wcBuf.data(), buf, n) == wxCONV_FAILED ||
2885 wxWcslen(wcBuf) != pszlen ||
2886 memcmp(wcBuf, psz, pszlen * sizeof(wchar_t)) != 0 )
2887 {
2888 // we didn't obtain the same thing we started from, hence
2889 // the conversion was lossy and we consider that it failed
2890 return wxCONV_FAILED;
2891 }
2892 }
2893
2894 return res ;
2895 }
2896
2897 virtual wxMBConv *Clone() const { return new wxMBConv_mac(*this); }
2898
2899 bool IsOk() const
2900 {
2901 CreateIfNeeded() ;
2902 return m_MB2WC_converter != NULL && m_WC2MB_converter != NULL;
2903 }
2904
2905protected :
2906 mutable TECObjectRef m_MB2WC_converter;
2907 mutable TECObjectRef m_WC2MB_converter;
2908
2909 TextEncodingBase m_char_encoding;
2910 TextEncodingBase m_unicode_encoding;
2911};
2912
2913// MB is decomposed (D) normalized UTF8
2914
2915class wxMBConv_macUTF8D : public wxMBConv_mac
2916{
2917public :
2918 wxMBConv_macUTF8D()
2919 {
2920 Init( kTextEncodingUnicodeDefault , kUnicodeNoSubset , kUnicodeUTF8Format ) ;
2921 m_uni = NULL;
2922 m_uniBack = NULL ;
2923 }
2924
2925 virtual ~wxMBConv_macUTF8D()
2926 {
2927 if (m_uni!=NULL)
2928 DisposeUnicodeToTextInfo(&m_uni);
2929 if (m_uniBack!=NULL)
2930 DisposeUnicodeToTextInfo(&m_uniBack);
2931 }
2932
2933 size_t WC2MB(char *buf, const wchar_t *psz, size_t n) const
2934 {
2935 CreateIfNeeded() ;
2936 OSStatus status = noErr ;
2937 ByteCount byteOutLen ;
2938 ByteCount byteInLen = wxWcslen(psz) * SIZEOF_WCHAR_T ;
2939
2940 char *tbuf = NULL ;
2941
2942 if (buf == NULL)
2943 {
2944 // Apple specs say at least 32
2945 n = wxMax( 32, ((byteInLen / SIZEOF_WCHAR_T) * 8) + SIZEOF_WCHAR_T );
2946 tbuf = (char*) malloc( n ) ;
2947 }
2948
2949 ByteCount byteBufferLen = n ;
2950 UniChar* ubuf = NULL ;
2951
2952#if SIZEOF_WCHAR_T == 4
2953 wxMBConvUTF16 converter ;
2954 size_t unicharlen = converter.WC2MB( NULL, psz, 0 ) ;
2955 byteInLen = unicharlen ;
2956 ubuf = (UniChar*) malloc( byteInLen + 2 ) ;
2957 converter.WC2MB( (char*) ubuf, psz, unicharlen + 2 ) ;
2958#else
2959 ubuf = (UniChar*) psz ;
2960#endif
2961
2962 // ubuf is a non-decomposed UniChar buffer
2963
2964 ByteCount dcubuflen = byteInLen * 2 + 2 ;
2965 ByteCount dcubufread , dcubufwritten ;
2966 UniChar *dcubuf = (UniChar*) malloc( dcubuflen ) ;
2967
2968 ConvertFromUnicodeToText( m_uni , byteInLen , ubuf ,
2969 kUnicodeDefaultDirectionMask, 0, NULL, NULL, NULL, dcubuflen , &dcubufread , &dcubufwritten , dcubuf ) ;
2970
2971 // we now convert that decomposed buffer into UTF8
2972
2973 status = TECConvertText(
2974 m_WC2MB_converter, (ConstTextPtr) dcubuf, dcubufwritten, &dcubufread,
2975 (TextPtr) (buf ? buf : tbuf), byteBufferLen, &byteOutLen);
2976
2977 free( dcubuf );
2978
2979#if SIZEOF_WCHAR_T == 4
2980 free( ubuf ) ;
2981#endif
2982
2983 if ( buf == NULL )
2984 free(tbuf) ;
2985
2986 size_t res = byteOutLen ;
2987 if ( buf && res < n)
2988 {
2989 buf[res] = 0;
2990 // don't test for round-trip fidelity yet, we cannot guarantee it yet
2991 }
2992
2993 return res ;
2994 }
2995
2996 size_t MB2WC(wchar_t *buf, const char *psz, size_t n) const
2997 {
2998 CreateIfNeeded() ;
2999 OSStatus status = noErr ;
3000 ByteCount byteOutLen ;
3001 ByteCount byteInLen = strlen(psz) + 1;
3002 wchar_t *tbuf = NULL ;
3003 UniChar* ubuf = NULL ;
3004 size_t res = 0 ;
3005
3006 if (buf == NULL)
3007 {
3008 // Apple specs say at least 32
3009 n = wxMax( 32, byteInLen ) ;
3010 tbuf = (wchar_t*) malloc( n * SIZEOF_WCHAR_T ) ;
3011 }
3012
3013 ByteCount byteBufferLen = n * sizeof( UniChar ) ;
3014
3015#if SIZEOF_WCHAR_T == 4
3016 ubuf = (UniChar*) malloc( byteBufferLen + 2 ) ;
3017#else
3018 ubuf = (UniChar*) (buf ? buf : tbuf) ;
3019#endif
3020
3021 ByteCount dcubuflen = byteBufferLen * 2 + 2 ;
3022 ByteCount dcubufread , dcubufwritten ;
3023 UniChar *dcubuf = (UniChar*) malloc( dcubuflen ) ;
3024
3025 status = TECConvertText(
3026 m_MB2WC_converter, (ConstTextPtr) psz, byteInLen, &byteInLen,
3027 (TextPtr) dcubuf, dcubuflen, &byteOutLen);
3028 // we have to terminate here, because n might be larger for the trailing zero, and if UniChar
3029 // is not properly terminated we get random characters at the end
3030 dcubuf[byteOutLen / sizeof( UniChar ) ] = 0 ;
3031
3032 // now from the decomposed UniChar to properly composed uniChar
3033 ConvertFromUnicodeToText( m_uniBack , byteOutLen , dcubuf ,
3034 kUnicodeDefaultDirectionMask, 0, NULL, NULL, NULL, dcubuflen , &dcubufread , &dcubufwritten , ubuf ) ;
3035
3036 free( dcubuf );
3037 byteOutLen = dcubufwritten ;
3038 ubuf[byteOutLen / sizeof( UniChar ) ] = 0 ;
3039
3040
3041#if SIZEOF_WCHAR_T == 4
3042 wxMBConvUTF16 converter ;
3043 res = converter.MB2WC( (buf ? buf : tbuf), (const char*)ubuf, n ) ;
3044 free( ubuf ) ;
3045#else
3046 res = byteOutLen / sizeof( UniChar ) ;
3047#endif
3048
3049 if ( buf == NULL )
3050 free(tbuf) ;
3051
3052 if ( buf && res < n)
3053 buf[res] = 0;
3054
3055 return res ;
3056 }
3057
3058 virtual void CreateIfNeeded() const
3059 {
3060 wxMBConv_mac::CreateIfNeeded() ;
3061 if ( m_uni == NULL )
3062 {
3063 m_map.unicodeEncoding = CreateTextEncoding(kTextEncodingUnicodeDefault,
3064 kUnicodeNoSubset, kTextEncodingDefaultFormat);
3065 m_map.otherEncoding = CreateTextEncoding(kTextEncodingUnicodeDefault,
3066 kUnicodeCanonicalDecompVariant, kTextEncodingDefaultFormat);
3067 m_map.mappingVersion = kUnicodeUseLatestMapping;
3068
3069 OSStatus err = CreateUnicodeToTextInfo(&m_map, &m_uni);
3070 wxASSERT_MSG( err == noErr , _(" Couldn't create the UnicodeConverter")) ;
3071
3072 m_map.unicodeEncoding = CreateTextEncoding(kTextEncodingUnicodeDefault,
3073 kUnicodeNoSubset, kTextEncodingDefaultFormat);
3074 m_map.otherEncoding = CreateTextEncoding(kTextEncodingUnicodeDefault,
3075 kUnicodeCanonicalCompVariant, kTextEncodingDefaultFormat);
3076 m_map.mappingVersion = kUnicodeUseLatestMapping;
3077 err = CreateUnicodeToTextInfo(&m_map, &m_uniBack);
3078 wxASSERT_MSG( err == noErr , _(" Couldn't create the UnicodeConverter")) ;
3079 }
3080 }
3081protected :
3082 mutable UnicodeToTextInfo m_uni;
3083 mutable UnicodeToTextInfo m_uniBack;
3084 mutable UnicodeMapping m_map;
3085};
3086#endif // defined(__WXMAC__) && defined(TARGET_CARBON)
3087
3088// ============================================================================
3089// wxEncodingConverter based conversion classes
3090// ============================================================================
3091
3092#if wxUSE_FONTMAP
3093
3094class wxMBConv_wxwin : public wxMBConv
3095{
3096private:
3097 void Init()
3098 {
3099 m_ok = m2w.Init(m_enc, wxFONTENCODING_UNICODE) &&
3100 w2m.Init(wxFONTENCODING_UNICODE, m_enc);
3101 }
3102
3103public:
3104 // temporarily just use wxEncodingConverter stuff,
3105 // so that it works while a better implementation is built
3106 wxMBConv_wxwin(const char* name)
3107 {
3108 if (name)
3109 m_enc = wxFontMapperBase::Get()->CharsetToEncoding(name, false);
3110 else
3111 m_enc = wxFONTENCODING_SYSTEM;
3112
3113 Init();
3114 }
3115
3116 wxMBConv_wxwin(wxFontEncoding enc)
3117 {
3118 m_enc = enc;
3119
3120 Init();
3121 }
3122
3123 size_t MB2WC(wchar_t *buf, const char *psz, size_t WXUNUSED(n)) const
3124 {
3125 size_t inbuf = strlen(psz);
3126 if (buf)
3127 {
3128 if (!m2w.Convert(psz, buf))
3129 return wxCONV_FAILED;
3130 }
3131 return inbuf;
3132 }
3133
3134 size_t WC2MB(char *buf, const wchar_t *psz, size_t WXUNUSED(n)) const
3135 {
3136 const size_t inbuf = wxWcslen(psz);
3137 if (buf)
3138 {
3139 if (!w2m.Convert(psz, buf))
3140 return wxCONV_FAILED;
3141 }
3142
3143 return inbuf;
3144 }
3145
3146 virtual size_t GetMBNulLen() const
3147 {
3148 switch ( m_enc )
3149 {
3150 case wxFONTENCODING_UTF16BE:
3151 case wxFONTENCODING_UTF16LE:
3152 return 2;
3153
3154 case wxFONTENCODING_UTF32BE:
3155 case wxFONTENCODING_UTF32LE:
3156 return 4;
3157
3158 default:
3159 return 1;
3160 }
3161 }
3162
3163 virtual wxMBConv *Clone() const { return new wxMBConv_wxwin(m_enc); }
3164
3165 bool IsOk() const { return m_ok; }
3166
3167public:
3168 wxFontEncoding m_enc;
3169 wxEncodingConverter m2w, w2m;
3170
3171private:
3172 // were we initialized successfully?
3173 bool m_ok;
3174
3175 DECLARE_NO_COPY_CLASS(wxMBConv_wxwin)
3176};
3177
3178// make the constructors available for unit testing
3179WXDLLIMPEXP_BASE wxMBConv* new_wxMBConv_wxwin( const char* name )
3180{
3181 wxMBConv_wxwin* result = new wxMBConv_wxwin( name );
3182 if ( !result->IsOk() )
3183 {
3184 delete result;
3185 return 0;
3186 }
3187
3188 return result;
3189}
3190
3191#endif // wxUSE_FONTMAP
3192
3193// ============================================================================
3194// wxCSConv implementation
3195// ============================================================================
3196
3197void wxCSConv::Init()
3198{
3199 m_name = NULL;
3200 m_convReal = NULL;
3201 m_deferred = true;
3202}
3203
3204wxCSConv::wxCSConv(const wxString& charset)
3205{
3206 Init();
3207
3208 if ( !charset.empty() )
3209 {
3210 SetName(charset.ToAscii());
3211 }
3212
3213#if wxUSE_FONTMAP
3214 m_encoding = wxFontMapperBase::GetEncodingFromName(charset);
3215#else
3216 m_encoding = wxFONTENCODING_SYSTEM;
3217#endif
3218}
3219
3220wxCSConv::wxCSConv(wxFontEncoding encoding)
3221{
3222 if ( encoding == wxFONTENCODING_MAX || encoding == wxFONTENCODING_DEFAULT )
3223 {
3224 wxFAIL_MSG( _T("invalid encoding value in wxCSConv ctor") );
3225
3226 encoding = wxFONTENCODING_SYSTEM;
3227 }
3228
3229 Init();
3230
3231 m_encoding = encoding;
3232}
3233
3234wxCSConv::~wxCSConv()
3235{
3236 Clear();
3237}
3238
3239wxCSConv::wxCSConv(const wxCSConv& conv)
3240 : wxMBConv()
3241{
3242 Init();
3243
3244 SetName(conv.m_name);
3245 m_encoding = conv.m_encoding;
3246}
3247
3248wxCSConv& wxCSConv::operator=(const wxCSConv& conv)
3249{
3250 Clear();
3251
3252 SetName(conv.m_name);
3253 m_encoding = conv.m_encoding;
3254
3255 return *this;
3256}
3257
3258void wxCSConv::Clear()
3259{
3260 free(m_name);
3261 delete m_convReal;
3262
3263 m_name = NULL;
3264 m_convReal = NULL;
3265}
3266
3267void wxCSConv::SetName(const char *charset)
3268{
3269 if (charset)
3270 {
3271 m_name = strdup(charset);
3272 m_deferred = true;
3273 }
3274}
3275
3276#if wxUSE_FONTMAP
3277
3278WX_DECLARE_HASH_MAP( wxFontEncoding, wxString, wxIntegerHash, wxIntegerEqual,
3279 wxEncodingNameCache );
3280
3281static wxEncodingNameCache gs_nameCache;
3282#endif
3283
3284wxMBConv *wxCSConv::DoCreate() const
3285{
3286#if wxUSE_FONTMAP
3287 wxLogTrace(TRACE_STRCONV,
3288 wxT("creating conversion for %s"),
3289 (m_name ? m_name
3290 : (const char*)wxFontMapperBase::GetEncodingName(m_encoding).mb_str()));
3291#endif // wxUSE_FONTMAP
3292
3293 // check for the special case of ASCII or ISO8859-1 charset: as we have
3294 // special knowledge of it anyhow, we don't need to create a special
3295 // conversion object
3296 if ( m_encoding == wxFONTENCODING_ISO8859_1 ||
3297 m_encoding == wxFONTENCODING_DEFAULT )
3298 {
3299 // don't convert at all
3300 return NULL;
3301 }
3302
3303 // we trust OS to do conversion better than we can so try external
3304 // conversion methods first
3305 //
3306 // the full order is:
3307 // 1. OS conversion (iconv() under Unix or Win32 API)
3308 // 2. hard coded conversions for UTF
3309 // 3. wxEncodingConverter as fall back
3310
3311 // step (1)
3312#ifdef HAVE_ICONV
3313#if !wxUSE_FONTMAP
3314 if ( m_name )
3315#endif // !wxUSE_FONTMAP
3316 {
3317#if wxUSE_FONTMAP
3318 wxFontEncoding encoding(m_encoding);
3319#endif
3320
3321 if ( m_name )
3322 {
3323 wxMBConv_iconv *conv = new wxMBConv_iconv(m_name);
3324 if ( conv->IsOk() )
3325 return conv;
3326
3327 delete conv;
3328
3329#if wxUSE_FONTMAP
3330 encoding =
3331 wxFontMapperBase::Get()->CharsetToEncoding(m_name, false);
3332#endif // wxUSE_FONTMAP
3333 }
3334#if wxUSE_FONTMAP
3335 {
3336 const wxEncodingNameCache::iterator it = gs_nameCache.find(encoding);
3337 if ( it != gs_nameCache.end() )
3338 {
3339 if ( it->second.empty() )
3340 return NULL;
3341
3342 wxMBConv_iconv *conv = new wxMBConv_iconv(it->second.ToAscii());
3343 if ( conv->IsOk() )
3344 return conv;
3345
3346 delete conv;
3347 }
3348
3349 const wxChar** names = wxFontMapperBase::GetAllEncodingNames(encoding);
3350 // CS : in case this does not return valid names (eg for MacRoman)
3351 // encoding got a 'failure' entry in the cache all the same,
3352 // although it just has to be created using a different method, so
3353 // only store failed iconv creation attempts (or perhaps we
3354 // shoulnd't do this at all ?)
3355 if ( names[0] != NULL )
3356 {
3357 for ( ; *names; ++names )
3358 {
3359 // FIXME-UTF8: wxFontMapperBase::GetAllEncodingNames()
3360 // will need changes that will obsolete this
3361 wxString name(*names);
3362 wxMBConv_iconv *conv = new wxMBConv_iconv(name.ToAscii());
3363 if ( conv->IsOk() )
3364 {
3365 gs_nameCache[encoding] = *names;
3366 return conv;
3367 }
3368
3369 delete conv;
3370 }
3371
3372 gs_nameCache[encoding] = _T(""); // cache the failure
3373 }
3374 }
3375#endif // wxUSE_FONTMAP
3376 }
3377#endif // HAVE_ICONV
3378
3379#ifdef wxHAVE_WIN32_MB2WC
3380 {
3381#if wxUSE_FONTMAP
3382 wxMBConv_win32 *conv = m_name ? new wxMBConv_win32(m_name)
3383 : new wxMBConv_win32(m_encoding);
3384 if ( conv->IsOk() )
3385 return conv;
3386
3387 delete conv;
3388#else
3389 return NULL;
3390#endif
3391 }
3392#endif // wxHAVE_WIN32_MB2WC
3393
3394#if defined(__WXMAC__)
3395 {
3396 // leave UTF16 and UTF32 to the built-ins of wx
3397 if ( m_name || ( m_encoding < wxFONTENCODING_UTF16BE ||
3398 ( m_encoding >= wxFONTENCODING_MACMIN && m_encoding <= wxFONTENCODING_MACMAX ) ) )
3399 {
3400#if wxUSE_FONTMAP
3401 wxMBConv_mac *conv = m_name ? new wxMBConv_mac(m_name)
3402 : new wxMBConv_mac(m_encoding);
3403#else
3404 wxMBConv_mac *conv = new wxMBConv_mac(m_encoding);
3405#endif
3406 if ( conv->IsOk() )
3407 return conv;
3408
3409 delete conv;
3410 }
3411 }
3412#endif
3413
3414#if defined(__WXCOCOA__)
3415 {
3416 if ( m_name || ( m_encoding <= wxFONTENCODING_UTF16 ) )
3417 {
3418#if wxUSE_FONTMAP
3419 wxMBConv_cocoa *conv = m_name ? new wxMBConv_cocoa(m_name)
3420 : new wxMBConv_cocoa(m_encoding);
3421#else
3422 wxMBConv_cocoa *conv = new wxMBConv_cocoa(m_encoding);
3423#endif
3424
3425 if ( conv->IsOk() )
3426 return conv;
3427
3428 delete conv;
3429 }
3430 }
3431#endif
3432 // step (2)
3433 wxFontEncoding enc = m_encoding;
3434#if wxUSE_FONTMAP
3435 if ( enc == wxFONTENCODING_SYSTEM && m_name )
3436 {
3437 // use "false" to suppress interactive dialogs -- we can be called from
3438 // anywhere and popping up a dialog from here is the last thing we want to
3439 // do
3440 enc = wxFontMapperBase::Get()->CharsetToEncoding(m_name, false);
3441 }
3442#endif // wxUSE_FONTMAP
3443
3444 switch ( enc )
3445 {
3446 case wxFONTENCODING_UTF7:
3447 return new wxMBConvUTF7;
3448
3449 case wxFONTENCODING_UTF8:
3450 return new wxMBConvUTF8;
3451
3452 case wxFONTENCODING_UTF16BE:
3453 return new wxMBConvUTF16BE;
3454
3455 case wxFONTENCODING_UTF16LE:
3456 return new wxMBConvUTF16LE;
3457
3458 case wxFONTENCODING_UTF32BE:
3459 return new wxMBConvUTF32BE;
3460
3461 case wxFONTENCODING_UTF32LE:
3462 return new wxMBConvUTF32LE;
3463
3464 default:
3465 // nothing to do but put here to suppress gcc warnings
3466 break;
3467 }
3468
3469 // step (3)
3470#if wxUSE_FONTMAP
3471 {
3472 wxMBConv_wxwin *conv = m_name ? new wxMBConv_wxwin(m_name)
3473 : new wxMBConv_wxwin(m_encoding);
3474 if ( conv->IsOk() )
3475 return conv;
3476
3477 delete conv;
3478 }
3479#endif // wxUSE_FONTMAP
3480
3481 // NB: This is a hack to prevent deadlock. What could otherwise happen
3482 // in Unicode build: wxConvLocal creation ends up being here
3483 // because of some failure and logs the error. But wxLog will try to
3484 // attach a timestamp, for which it will need wxConvLocal (to convert
3485 // time to char* and then wchar_t*), but that fails, tries to log the
3486 // error, but wxLog has an (already locked) critical section that
3487 // guards the static buffer.
3488 static bool alreadyLoggingError = false;
3489 if (!alreadyLoggingError)
3490 {
3491 alreadyLoggingError = true;
3492 wxLogError(_("Cannot convert from the charset '%s'!"),
3493 m_name ? m_name
3494 :
3495#if wxUSE_FONTMAP
3496 (const char*)wxFontMapperBase::GetEncodingDescription(m_encoding).ToAscii()
3497#else // !wxUSE_FONTMAP
3498 (const char*)wxString::Format(_("encoding %i"), m_encoding).ToAscii()
3499#endif // wxUSE_FONTMAP/!wxUSE_FONTMAP
3500 );
3501
3502 alreadyLoggingError = false;
3503 }
3504
3505 return NULL;
3506}
3507
3508void wxCSConv::CreateConvIfNeeded() const
3509{
3510 if ( m_deferred )
3511 {
3512 wxCSConv *self = (wxCSConv *)this; // const_cast
3513
3514 // if we don't have neither the name nor the encoding, use the default
3515 // encoding for this system
3516 if ( !m_name && m_encoding == wxFONTENCODING_SYSTEM )
3517 {
3518#if wxUSE_INTL
3519 self->m_encoding = wxLocale::GetSystemEncoding();
3520#else
3521 // fallback to some reasonable default:
3522 self->m_encoding = wxFONTENCODING_ISO8859_1;
3523#endif // wxUSE_INTL
3524 }
3525
3526 self->m_convReal = DoCreate();
3527 self->m_deferred = false;
3528 }
3529}
3530
3531bool wxCSConv::IsOk() const
3532{
3533 CreateConvIfNeeded();
3534
3535 // special case: no convReal created for wxFONTENCODING_ISO8859_1
3536 if ( m_encoding == wxFONTENCODING_ISO8859_1 )
3537 return true; // always ok as we do it ourselves
3538
3539 // m_convReal->IsOk() is called at its own creation, so we know it must
3540 // be ok if m_convReal is non-NULL
3541 return m_convReal != NULL;
3542}
3543
3544size_t wxCSConv::ToWChar(wchar_t *dst, size_t dstLen,
3545 const char *src, size_t srcLen) const
3546{
3547 CreateConvIfNeeded();
3548
3549 if (m_convReal)
3550 return m_convReal->ToWChar(dst, dstLen, src, srcLen);
3551
3552 // latin-1 (direct)
3553 return wxMBConv::ToWChar(dst, dstLen, src, srcLen);
3554}
3555
3556size_t wxCSConv::FromWChar(char *dst, size_t dstLen,
3557 const wchar_t *src, size_t srcLen) const
3558{
3559 CreateConvIfNeeded();
3560
3561 if (m_convReal)
3562 return m_convReal->FromWChar(dst, dstLen, src, srcLen);
3563
3564 // latin-1 (direct)
3565 return wxMBConv::FromWChar(dst, dstLen, src, srcLen);
3566}
3567
3568size_t wxCSConv::MB2WC(wchar_t *buf, const char *psz, size_t n) const
3569{
3570 CreateConvIfNeeded();
3571
3572 if (m_convReal)
3573 return m_convReal->MB2WC(buf, psz, n);
3574
3575 // latin-1 (direct)
3576 size_t len = strlen(psz);
3577
3578 if (buf)
3579 {
3580 for (size_t c = 0; c <= len; c++)
3581 buf[c] = (unsigned char)(psz[c]);
3582 }
3583
3584 return len;
3585}
3586
3587size_t wxCSConv::WC2MB(char *buf, const wchar_t *psz, size_t n) const
3588{
3589 CreateConvIfNeeded();
3590
3591 if (m_convReal)
3592 return m_convReal->WC2MB(buf, psz, n);
3593
3594 // latin-1 (direct)
3595 const size_t len = wxWcslen(psz);
3596 if (buf)
3597 {
3598 for (size_t c = 0; c <= len; c++)
3599 {
3600 if (psz[c] > 0xFF)
3601 return wxCONV_FAILED;
3602
3603 buf[c] = (char)psz[c];
3604 }
3605 }
3606 else
3607 {
3608 for (size_t c = 0; c <= len; c++)
3609 {
3610 if (psz[c] > 0xFF)
3611 return wxCONV_FAILED;
3612 }
3613 }
3614
3615 return len;
3616}
3617
3618size_t wxCSConv::GetMBNulLen() const
3619{
3620 CreateConvIfNeeded();
3621
3622 if ( m_convReal )
3623 {
3624 return m_convReal->GetMBNulLen();
3625 }
3626
3627 // otherwise, we are ISO-8859-1
3628 return 1;
3629}
3630
3631#if wxUSE_UNICODE_UTF8
3632bool wxCSConv::IsUTF8() const
3633{
3634 CreateConvIfNeeded();
3635
3636 if ( m_convReal )
3637 {
3638 return m_convReal->IsUTF8();
3639 }
3640
3641 // otherwise, we are ISO-8859-1
3642 return false;
3643}
3644#endif
3645
3646
3647#if wxUSE_UNICODE
3648
3649wxWCharBuffer wxSafeConvertMB2WX(const char *s)
3650{
3651 if ( !s )
3652 return wxWCharBuffer();
3653
3654 wxWCharBuffer wbuf(wxConvLibc.cMB2WX(s));
3655 if ( !wbuf )
3656 wbuf = wxMBConvUTF8().cMB2WX(s);
3657 if ( !wbuf )
3658 wbuf = wxConvISO8859_1.cMB2WX(s);
3659
3660 return wbuf;
3661}
3662
3663wxCharBuffer wxSafeConvertWX2MB(const wchar_t *ws)
3664{
3665 if ( !ws )
3666 return wxCharBuffer();
3667
3668 wxCharBuffer buf(wxConvLibc.cWX2MB(ws));
3669 if ( !buf )
3670 buf = wxMBConvUTF8(wxMBConvUTF8::MAP_INVALID_UTF8_TO_OCTAL).cWX2MB(ws);
3671
3672 return buf;
3673}
3674
3675#endif // wxUSE_UNICODE
3676
3677// ----------------------------------------------------------------------------
3678// globals
3679// ----------------------------------------------------------------------------
3680
3681// NB: The reason why we create converted objects in this convoluted way,
3682// using a factory function instead of global variable, is that they
3683// may be used at static initialization time (some of them are used by
3684// wxString ctors and there may be a global wxString object). In other
3685// words, possibly _before_ the converter global object would be
3686// initialized.
3687
3688#undef wxConvLibc
3689#undef wxConvUTF8
3690#undef wxConvUTF7
3691#undef wxConvLocal
3692#undef wxConvISO8859_1
3693
3694#define WX_DEFINE_GLOBAL_CONV2(klass, impl_klass, name, ctor_args) \
3695 WXDLLIMPEXP_DATA_BASE(klass*) name##Ptr = NULL; \
3696 WXDLLIMPEXP_BASE klass* wxGet_##name##Ptr() \
3697 { \
3698 static impl_klass name##Obj ctor_args; \
3699 return &name##Obj; \
3700 } \
3701 /* this ensures that all global converter objects are created */ \
3702 /* by the time static initialization is done, i.e. before any */ \
3703 /* thread is launched: */ \
3704 static klass* gs_##name##instance = wxGet_##name##Ptr()
3705
3706#define WX_DEFINE_GLOBAL_CONV(klass, name, ctor_args) \
3707 WX_DEFINE_GLOBAL_CONV2(klass, klass, name, ctor_args)
3708
3709#ifdef __WINDOWS__
3710 WX_DEFINE_GLOBAL_CONV2(wxMBConv, wxMBConv_win32, wxConvLibc, wxEMPTY_PARAMETER_VALUE);
3711#elif defined(__WXMAC__) && !defined(__MACH__)
3712 WX_DEFINE_GLOBAL_CONV2(wxMBConv, wxMBConv_mac, wxConvLibc, wxEMPTY_PARAMETER_VALUE);
3713#else
3714 WX_DEFINE_GLOBAL_CONV2(wxMBConv, wxMBConvLibc, wxConvLibc, wxEMPTY_PARAMETER_VALUE);
3715#endif
3716
3717WX_DEFINE_GLOBAL_CONV(wxMBConvUTF8, wxConvUTF8, wxEMPTY_PARAMETER_VALUE);
3718WX_DEFINE_GLOBAL_CONV(wxMBConvUTF7, wxConvUTF7, wxEMPTY_PARAMETER_VALUE);
3719
3720WX_DEFINE_GLOBAL_CONV(wxCSConv, wxConvLocal, (wxFONTENCODING_SYSTEM));
3721WX_DEFINE_GLOBAL_CONV(wxCSConv, wxConvISO8859_1, (wxFONTENCODING_ISO8859_1));
3722
3723WXDLLIMPEXP_DATA_BASE(wxMBConv *) wxConvCurrent = wxGet_wxConvLibcPtr();
3724WXDLLIMPEXP_DATA_BASE(wxMBConv *) wxConvUI = wxGet_wxConvLocalPtr();
3725
3726#if defined(__WXMAC__) && defined(TARGET_CARBON)
3727static wxMBConv_macUTF8D wxConvMacUTF8DObj;
3728#endif
3729WXDLLIMPEXP_DATA_BASE(wxMBConv *) wxConvFileName =
3730#ifdef __WXOSX__
3731#if defined(__WXMAC__) && defined(TARGET_CARBON)
3732 &wxConvMacUTF8DObj;
3733#else
3734 wxGet_wxConvUTF8Ptr();
3735#endif
3736#else // !__WXOSX__
3737 wxGet_wxConvLibcPtr();
3738#endif // __WXOSX__/!__WXOSX__
3739
3740#else // !wxUSE_WCHAR_T
3741
3742// FIXME-UTF8: remove this, wxUSE_WCHAR_T is required now
3743// stand-ins in absence of wchar_t
3744WXDLLIMPEXP_DATA_BASE(wxMBConv) wxConvLibc,
3745 wxConvISO8859_1,
3746 wxConvLocal,
3747 wxConvUTF8;
3748
3749#endif // wxUSE_WCHAR_T/!wxUSE_WCHAR_T
wxWidgets (Impactor)