]> git.saurik.com Git - wxWidgets.git/blame - src/common/strconv.cpp
cleanup image after application
[wxWidgets.git] / src / common / strconv.cpp
CommitLineData
6001e347
RR
1/////////////////////////////////////////////////////////////////////////////
2// Name: strconv.cpp
3// Purpose: Unicode conversion classes
3a0d76bc 4// Author: Ove Kaaven, Robert Roebling, Vadim Zeitlin, Vaclav Slavik
6001e347
RR
5// Modified by:
6// Created: 29/01/98
7// RCS-ID: $Id$
e95354ec
VZ
8// Copyright: (c) 1999 Ove Kaaven, Robert Roebling, Vaclav Slavik
9// (c) 2000-2003 Vadim Zeitlin
65571936 10// Licence: wxWindows licence
6001e347
RR
11/////////////////////////////////////////////////////////////////////////////
12
f6bcfd97
BP
13// ============================================================================
14// declarations
15// ============================================================================
16
17// ----------------------------------------------------------------------------
18// headers
19// ----------------------------------------------------------------------------
20
14f355c2 21#if defined(__GNUG__) && !defined(NO_GCC_PRAGMA)
6001e347
RR
22 #pragma implementation "strconv.h"
23#endif
24
25// For compilers that support precompilation, includes "wx.h".
26#include "wx/wxprec.h"
27
28#ifdef __BORLANDC__
29 #pragma hdrstop
30#endif
31
373658eb
VZ
32#ifndef WX_PRECOMP
33 #include "wx/intl.h"
34 #include "wx/log.h"
35#endif // WX_PRECOMP
36
bde4baac
VZ
37#include "wx/strconv.h"
38
39#if wxUSE_WCHAR_T
40
0a1c1e62 41#ifdef __WXMSW__
373658eb 42 #include "wx/msw/private.h"
7608a683
WS
43#endif
44
45#ifdef __WINDOWS__
13dd924a 46 #include "wx/msw/missing.h"
0a1c1e62
GRG
47#endif
48
1c193821 49#ifndef __WXWINCE__
1cd52418 50#include <errno.h>
1c193821
JS
51#endif
52
6001e347
RR
53#include <ctype.h>
54#include <string.h>
55#include <stdlib.h>
56
e95354ec
VZ
57#if defined(__WIN32__) && !defined(__WXMICROWIN__)
58 #define wxHAVE_WIN32_MB2WC
59#endif // __WIN32__ but !__WXMICROWIN__
60
373658eb
VZ
61// ----------------------------------------------------------------------------
62// headers
63// ----------------------------------------------------------------------------
7af284fd 64
6001e347 65#ifdef __SALFORDC__
373658eb 66 #include <clib.h>
6001e347
RR
67#endif
68
b040e242 69#ifdef HAVE_ICONV
373658eb 70 #include <iconv.h>
1cd52418 71#endif
1cd52418 72
373658eb
VZ
73#include "wx/encconv.h"
74#include "wx/fontmap.h"
7608a683 75#include "wx/utils.h"
373658eb 76
335d31e0 77#ifdef __WXMAC__
4227afa4
SC
78#include <ATSUnicode.h>
79#include <TextCommon.h>
80#include <TextEncodingConverter.h>
335d31e0
SC
81
82#include "wx/mac/private.h" // includes mac headers
83#endif
373658eb
VZ
84// ----------------------------------------------------------------------------
85// macros
86// ----------------------------------------------------------------------------
3e61dfb0 87
1cd52418 88#define BSWAP_UCS4(str, len) { unsigned _c; for (_c=0; _c<len; _c++) str[_c]=wxUINT32_SWAP_ALWAYS(str[_c]); }
3a0d76bc 89#define BSWAP_UTF16(str, len) { unsigned _c; for (_c=0; _c<len; _c++) str[_c]=wxUINT16_SWAP_ALWAYS(str[_c]); }
1cd52418
OK
90
91#if SIZEOF_WCHAR_T == 4
3a0d76bc
VS
92 #define WC_NAME "UCS4"
93 #define WC_BSWAP BSWAP_UCS4
94 #ifdef WORDS_BIGENDIAN
95 #define WC_NAME_BEST "UCS-4BE"
96 #else
97 #define WC_NAME_BEST "UCS-4LE"
98 #endif
1cd52418 99#elif SIZEOF_WCHAR_T == 2
3a0d76bc
VS
100 #define WC_NAME "UTF16"
101 #define WC_BSWAP BSWAP_UTF16
a3f2769e 102 #define WC_UTF16
3a0d76bc
VS
103 #ifdef WORDS_BIGENDIAN
104 #define WC_NAME_BEST "UTF-16BE"
105 #else
106 #define WC_NAME_BEST "UTF-16LE"
107 #endif
bab1e722 108#else // sizeof(wchar_t) != 2 nor 4
bde4baac
VZ
109 // does this ever happen?
110 #error "Unknown sizeof(wchar_t): please report this to wx-dev@lists.wxwindows.org"
1cd52418
OK
111#endif
112
373658eb
VZ
113// ============================================================================
114// implementation
115// ============================================================================
116
117// ----------------------------------------------------------------------------
c91830cb 118// UTF-16 en/decoding to/from UCS-4
373658eb 119// ----------------------------------------------------------------------------
6001e347 120
b0a6bb75 121
c91830cb 122static size_t encode_utf16(wxUint32 input, wxUint16 *output)
1cd52418 123{
dccce9ea 124 if (input<=0xffff)
4def3b35 125 {
999836aa
VZ
126 if (output)
127 *output = (wxUint16) input;
4def3b35 128 return 1;
dccce9ea
VZ
129 }
130 else if (input>=0x110000)
4def3b35
VS
131 {
132 return (size_t)-1;
dccce9ea
VZ
133 }
134 else
4def3b35 135 {
dccce9ea 136 if (output)
4def3b35 137 {
c91830cb 138 *output++ = (wxUint16) ((input >> 10)+0xd7c0);
999836aa 139 *output = (wxUint16) ((input&0x3ff)+0xdc00);
4def3b35
VS
140 }
141 return 2;
1cd52418 142 }
1cd52418
OK
143}
144
c91830cb 145static size_t decode_utf16(const wxUint16* input, wxUint32& output)
1cd52418 146{
dccce9ea 147 if ((*input<0xd800) || (*input>0xdfff))
4def3b35
VS
148 {
149 output = *input;
150 return 1;
dccce9ea
VZ
151 }
152 else if ((input[1]<0xdc00) || (input[1]>=0xdfff))
4def3b35
VS
153 {
154 output = *input;
155 return (size_t)-1;
dccce9ea
VZ
156 }
157 else
4def3b35
VS
158 {
159 output = ((input[0] - 0xd7c0) << 10) + (input[1] - 0xdc00);
160 return 2;
161 }
1cd52418
OK
162}
163
b0a6bb75 164
f6bcfd97 165// ----------------------------------------------------------------------------
6001e347 166// wxMBConv
f6bcfd97 167// ----------------------------------------------------------------------------
6001e347 168
2b5f62a0
VZ
169wxMBConv::~wxMBConv()
170{
171 // nothing to do here
172}
173
6001e347
RR
174const wxWCharBuffer wxMBConv::cMB2WC(const char *psz) const
175{
2b5f62a0 176 if ( psz )
6001e347 177 {
2b5f62a0
VZ
178 // calculate the length of the buffer needed first
179 size_t nLen = MB2WC(NULL, psz, 0);
180 if ( nLen != (size_t)-1 )
181 {
182 // now do the actual conversion
183 wxWCharBuffer buf(nLen);
635f33ce
VS
184 nLen = MB2WC(buf.data(), psz, nLen + 1); // with the trailing NULL
185 if ( nLen != (size_t)-1 )
186 {
187 return buf;
188 }
2b5f62a0 189 }
f6bcfd97 190 }
2b5f62a0
VZ
191
192 wxWCharBuffer buf((wchar_t *)NULL);
193
194 return buf;
6001e347
RR
195}
196
e5cceba0 197const wxCharBuffer wxMBConv::cWC2MB(const wchar_t *pwz) const
6001e347 198{
2b5f62a0
VZ
199 if ( pwz )
200 {
201 size_t nLen = WC2MB(NULL, pwz, 0);
202 if ( nLen != (size_t)-1 )
203 {
c91830cb 204 wxCharBuffer buf(nLen+3); // space for a wxUint32 trailing zero
635f33ce
VS
205 nLen = WC2MB(buf.data(), pwz, nLen + 4);
206 if ( nLen != (size_t)-1 )
207 {
208 return buf;
209 }
2b5f62a0
VZ
210 }
211 }
212
213 wxCharBuffer buf((char *)NULL);
e5cceba0 214
e5cceba0 215 return buf;
6001e347
RR
216}
217
6001e347 218// ----------------------------------------------------------------------------
bde4baac 219// wxMBConvLibc
6001e347
RR
220// ----------------------------------------------------------------------------
221
bde4baac
VZ
222size_t wxMBConvLibc::MB2WC(wchar_t *buf, const char *psz, size_t n) const
223{
224 return wxMB2WC(buf, psz, n);
225}
226
227size_t wxMBConvLibc::WC2MB(char *buf, const wchar_t *psz, size_t n) const
228{
229 return wxWC2MB(buf, psz, n);
230}
231
232// ----------------------------------------------------------------------------
233// UTF-7
234// ----------------------------------------------------------------------------
6001e347
RR
235
236#if 0
237static char utf7_setD[]="ABCDEFGHIJKLMNOPQRSTUVWXYZ"
238 "abcdefghijklmnopqrstuvwxyz"
239 "0123456789'(),-./:?";
240static char utf7_setO[]="!\"#$%&*;<=>@[]^_`{|}";
241static char utf7_setB[]="ABCDEFGHIJKLMNOPQRSTUVWXYZ"
242 "abcdefghijklmnopqrstuvwxyz"
243 "0123456789+/";
244#endif
245
246// TODO: write actual implementations of UTF-7 here
247size_t wxMBConvUTF7::MB2WC(wchar_t * WXUNUSED(buf),
248 const char * WXUNUSED(psz),
249 size_t WXUNUSED(n)) const
250{
251 return 0;
252}
253
254size_t wxMBConvUTF7::WC2MB(char * WXUNUSED(buf),
255 const wchar_t * WXUNUSED(psz),
256 size_t WXUNUSED(n)) const
257{
258 return 0;
259}
260
f6bcfd97 261// ----------------------------------------------------------------------------
6001e347 262// UTF-8
f6bcfd97 263// ----------------------------------------------------------------------------
6001e347 264
dccce9ea 265static wxUint32 utf8_max[]=
4def3b35 266 { 0x7f, 0x7ff, 0xffff, 0x1fffff, 0x3ffffff, 0x7fffffff, 0xffffffff };
6001e347
RR
267
268size_t wxMBConvUTF8::MB2WC(wchar_t *buf, const char *psz, size_t n) const
269{
4def3b35
VS
270 size_t len = 0;
271
dccce9ea 272 while (*psz && ((!buf) || (len < n)))
4def3b35
VS
273 {
274 unsigned char cc = *psz++, fc = cc;
275 unsigned cnt;
dccce9ea 276 for (cnt = 0; fc & 0x80; cnt++)
4def3b35 277 fc <<= 1;
dccce9ea 278 if (!cnt)
4def3b35
VS
279 {
280 // plain ASCII char
dccce9ea 281 if (buf)
4def3b35
VS
282 *buf++ = cc;
283 len++;
dccce9ea
VZ
284 }
285 else
4def3b35
VS
286 {
287 cnt--;
dccce9ea 288 if (!cnt)
4def3b35
VS
289 {
290 // invalid UTF-8 sequence
291 return (size_t)-1;
dccce9ea
VZ
292 }
293 else
4def3b35
VS
294 {
295 unsigned ocnt = cnt - 1;
296 wxUint32 res = cc & (0x3f >> cnt);
dccce9ea 297 while (cnt--)
4def3b35
VS
298 {
299 cc = *psz++;
dccce9ea 300 if ((cc & 0xC0) != 0x80)
4def3b35
VS
301 {
302 // invalid UTF-8 sequence
303 return (size_t)-1;
304 }
305 res = (res << 6) | (cc & 0x3f);
306 }
dccce9ea 307 if (res <= utf8_max[ocnt])
4def3b35
VS
308 {
309 // illegal UTF-8 encoding
310 return (size_t)-1;
311 }
1cd52418 312#ifdef WC_UTF16
b5153fd8
VZ
313 // cast is ok because wchar_t == wxUuint16 if WC_UTF16
314 size_t pa = encode_utf16(res, (wxUint16 *)buf);
4def3b35
VS
315 if (pa == (size_t)-1)
316 return (size_t)-1;
dccce9ea 317 if (buf)
4def3b35
VS
318 buf += pa;
319 len += pa;
373658eb 320#else // !WC_UTF16
dccce9ea 321 if (buf)
4def3b35
VS
322 *buf++ = res;
323 len++;
373658eb 324#endif // WC_UTF16/!WC_UTF16
4def3b35
VS
325 }
326 }
6001e347 327 }
dccce9ea 328 if (buf && (len < n))
4def3b35
VS
329 *buf = 0;
330 return len;
6001e347
RR
331}
332
333size_t wxMBConvUTF8::WC2MB(char *buf, const wchar_t *psz, size_t n) const
334{
4def3b35 335 size_t len = 0;
6001e347 336
dccce9ea 337 while (*psz && ((!buf) || (len < n)))
4def3b35
VS
338 {
339 wxUint32 cc;
1cd52418 340#ifdef WC_UTF16
b5153fd8
VZ
341 // cast is ok for WC_UTF16
342 size_t pa = decode_utf16((const wxUint16 *)psz, cc);
4def3b35 343 psz += (pa == (size_t)-1) ? 1 : pa;
1cd52418 344#else
4def3b35
VS
345 cc=(*psz++) & 0x7fffffff;
346#endif
347 unsigned cnt;
348 for (cnt = 0; cc > utf8_max[cnt]; cnt++) {}
dccce9ea 349 if (!cnt)
4def3b35
VS
350 {
351 // plain ASCII char
dccce9ea 352 if (buf)
574c939e 353 *buf++ = (char) cc;
4def3b35 354 len++;
dccce9ea
VZ
355 }
356
357 else
4def3b35
VS
358 {
359 len += cnt + 1;
dccce9ea 360 if (buf)
4def3b35 361 {
574c939e 362 *buf++ = (char) ((-128 >> cnt) | ((cc >> (cnt * 6)) & (0x3f >> cnt)));
4def3b35 363 while (cnt--)
574c939e 364 *buf++ = (char) (0x80 | ((cc >> (cnt * 6)) & 0x3f));
4def3b35
VS
365 }
366 }
6001e347 367 }
4def3b35
VS
368
369 if (buf && (len<n)) *buf = 0;
adb45366 370
4def3b35 371 return len;
6001e347
RR
372}
373
c91830cb
VZ
374
375
376
377// ----------------------------------------------------------------------------
378// UTF-16
379// ----------------------------------------------------------------------------
380
381#ifdef WORDS_BIGENDIAN
bde4baac
VZ
382 #define wxMBConvUTF16straight wxMBConvUTF16BE
383 #define wxMBConvUTF16swap wxMBConvUTF16LE
c91830cb 384#else
bde4baac
VZ
385 #define wxMBConvUTF16swap wxMBConvUTF16BE
386 #define wxMBConvUTF16straight wxMBConvUTF16LE
c91830cb
VZ
387#endif
388
389
c91830cb
VZ
390#ifdef WC_UTF16
391
c91830cb
VZ
392// copy 16bit MB to 16bit String
393size_t wxMBConvUTF16straight::MB2WC(wchar_t *buf, const char *psz, size_t n) const
394{
395 size_t len=0;
396
397 while (*(wxUint16*)psz && (!buf || len < n))
398 {
399 if (buf)
400 *buf++ = *(wxUint16*)psz;
401 len++;
402
403 psz += sizeof(wxUint16);
404 }
405 if (buf && len<n) *buf=0;
406
407 return len;
408}
409
410
411// copy 16bit String to 16bit MB
412size_t wxMBConvUTF16straight::WC2MB(char *buf, const wchar_t *psz, size_t n) const
413{
414 size_t len=0;
415
416 while (*psz && (!buf || len < n))
417 {
418 if (buf)
419 {
420 *(wxUint16*)buf = *psz;
421 buf += sizeof(wxUint16);
422 }
423 len += sizeof(wxUint16);
424 psz++;
425 }
426 if (buf && len<=n-sizeof(wxUint16)) *(wxUint16*)buf=0;
427
428 return len;
429}
430
431
432// swap 16bit MB to 16bit String
433size_t wxMBConvUTF16swap::MB2WC(wchar_t *buf, const char *psz, size_t n) const
434{
435 size_t len=0;
436
437 while (*(wxUint16*)psz && (!buf || len < n))
438 {
439 if (buf)
440 {
441 ((char *)buf)[0] = psz[1];
442 ((char *)buf)[1] = psz[0];
443 buf++;
444 }
445 len++;
446 psz += sizeof(wxUint16);
447 }
448 if (buf && len<n) *buf=0;
449
450 return len;
451}
452
453
454// swap 16bit MB to 16bit String
455size_t wxMBConvUTF16swap::WC2MB(char *buf, const wchar_t *psz, size_t n) const
456{
457 size_t len=0;
458
459 while (*psz && (!buf || len < n))
460 {
461 if (buf)
462 {
463 *buf++ = ((char*)psz)[1];
464 *buf++ = ((char*)psz)[0];
465 }
466 len += sizeof(wxUint16);
467 psz++;
468 }
469 if (buf && len<=n-sizeof(wxUint16)) *(wxUint16*)buf=0;
470
471 return len;
472}
473
474
475#else // WC_UTF16
476
477
478// copy 16bit MB to 32bit String
479size_t wxMBConvUTF16straight::MB2WC(wchar_t *buf, const char *psz, size_t n) const
480{
481 size_t len=0;
482
483 while (*(wxUint16*)psz && (!buf || len < n))
484 {
485 wxUint32 cc;
486 size_t pa=decode_utf16((wxUint16*)psz, cc);
487 if (pa == (size_t)-1)
488 return pa;
489
490 if (buf)
491 *buf++ = cc;
492 len++;
493 psz += pa * sizeof(wxUint16);
494 }
495 if (buf && len<n) *buf=0;
496
497 return len;
498}
499
500
501// copy 32bit String to 16bit MB
502size_t wxMBConvUTF16straight::WC2MB(char *buf, const wchar_t *psz, size_t n) const
503{
504 size_t len=0;
505
506 while (*psz && (!buf || len < n))
507 {
508 wxUint16 cc[2];
509 size_t pa=encode_utf16(*psz, cc);
510
511 if (pa == (size_t)-1)
512 return pa;
513
514 if (buf)
515 {
69b80d28 516 *(wxUint16*)buf = cc[0];
b5153fd8 517 buf += sizeof(wxUint16);
c91830cb 518 if (pa > 1)
69b80d28
VZ
519 {
520 *(wxUint16*)buf = cc[1];
521 buf += sizeof(wxUint16);
522 }
c91830cb
VZ
523 }
524
525 len += pa*sizeof(wxUint16);
526 psz++;
527 }
528 if (buf && len<=n-sizeof(wxUint16)) *(wxUint16*)buf=0;
529
530 return len;
531}
532
533
534// swap 16bit MB to 32bit String
535size_t wxMBConvUTF16swap::MB2WC(wchar_t *buf, const char *psz, size_t n) const
536{
537 size_t len=0;
538
539 while (*(wxUint16*)psz && (!buf || len < n))
540 {
541 wxUint32 cc;
542 char tmp[4];
543 tmp[0]=psz[1]; tmp[1]=psz[0];
544 tmp[2]=psz[3]; tmp[3]=psz[2];
545
546 size_t pa=decode_utf16((wxUint16*)tmp, cc);
547 if (pa == (size_t)-1)
548 return pa;
549
550 if (buf)
551 *buf++ = cc;
552
553 len++;
554 psz += pa * sizeof(wxUint16);
555 }
556 if (buf && len<n) *buf=0;
557
558 return len;
559}
560
561
562// swap 32bit String to 16bit MB
563size_t wxMBConvUTF16swap::WC2MB(char *buf, const wchar_t *psz, size_t n) const
564{
565 size_t len=0;
566
567 while (*psz && (!buf || len < n))
568 {
569 wxUint16 cc[2];
570 size_t pa=encode_utf16(*psz, cc);
571
572 if (pa == (size_t)-1)
573 return pa;
574
575 if (buf)
576 {
577 *buf++ = ((char*)cc)[1];
578 *buf++ = ((char*)cc)[0];
579 if (pa > 1)
580 {
581 *buf++ = ((char*)cc)[3];
582 *buf++ = ((char*)cc)[2];
583 }
584 }
585
586 len += pa*sizeof(wxUint16);
587 psz++;
588 }
589 if (buf && len<=n-sizeof(wxUint16)) *(wxUint16*)buf=0;
590
591 return len;
592}
593
594#endif // WC_UTF16
595
596
597// ----------------------------------------------------------------------------
598// UTF-32
599// ----------------------------------------------------------------------------
600
601#ifdef WORDS_BIGENDIAN
602#define wxMBConvUTF32straight wxMBConvUTF32BE
603#define wxMBConvUTF32swap wxMBConvUTF32LE
604#else
605#define wxMBConvUTF32swap wxMBConvUTF32BE
606#define wxMBConvUTF32straight wxMBConvUTF32LE
607#endif
608
609
610WXDLLIMPEXP_DATA_BASE(wxMBConvUTF32LE) wxConvUTF32LE;
611WXDLLIMPEXP_DATA_BASE(wxMBConvUTF32BE) wxConvUTF32BE;
612
613
614#ifdef WC_UTF16
615
616// copy 32bit MB to 16bit String
617size_t wxMBConvUTF32straight::MB2WC(wchar_t *buf, const char *psz, size_t n) const
618{
619 size_t len=0;
620
621 while (*(wxUint32*)psz && (!buf || len < n))
622 {
623 wxUint16 cc[2];
624
625 size_t pa=encode_utf16(*(wxUint32*)psz, cc);
626 if (pa == (size_t)-1)
627 return pa;
628
629 if (buf)
630 {
631 *buf++ = cc[0];
632 if (pa > 1)
633 *buf++ = cc[1];
634 }
635 len += pa;
636 psz += sizeof(wxUint32);
637 }
638 if (buf && len<n) *buf=0;
639
640 return len;
641}
642
643
644// copy 16bit String to 32bit MB
645size_t wxMBConvUTF32straight::WC2MB(char *buf, const wchar_t *psz, size_t n) const
646{
647 size_t len=0;
648
649 while (*psz && (!buf || len < n))
650 {
651 wxUint32 cc;
652
b5153fd8
VZ
653 // cast is ok for WC_UTF16
654 size_t pa = decode_utf16((const wxUint16 *)psz, cc);
c91830cb
VZ
655 if (pa == (size_t)-1)
656 return pa;
657
658 if (buf)
659 {
660 *(wxUint32*)buf = cc;
661 buf += sizeof(wxUint32);
662 }
663 len += sizeof(wxUint32);
664 psz += pa;
665 }
b5153fd8
VZ
666
667 if (buf && len<=n-sizeof(wxUint32))
668 *(wxUint32*)buf=0;
c91830cb
VZ
669
670 return len;
671}
672
673
674
675// swap 32bit MB to 16bit String
676size_t wxMBConvUTF32swap::MB2WC(wchar_t *buf, const char *psz, size_t n) const
677{
678 size_t len=0;
679
680 while (*(wxUint32*)psz && (!buf || len < n))
681 {
682 char tmp[4];
683 tmp[0] = psz[3]; tmp[1] = psz[2];
684 tmp[2] = psz[1]; tmp[3] = psz[0];
685
686
687 wxUint16 cc[2];
688
689 size_t pa=encode_utf16(*(wxUint32*)tmp, cc);
690 if (pa == (size_t)-1)
691 return pa;
692
693 if (buf)
694 {
695 *buf++ = cc[0];
696 if (pa > 1)
697 *buf++ = cc[1];
698 }
699 len += pa;
700 psz += sizeof(wxUint32);
701 }
b5153fd8
VZ
702
703 if (buf && len<n)
704 *buf=0;
c91830cb
VZ
705
706 return len;
707}
708
709
710// swap 16bit String to 32bit MB
711size_t wxMBConvUTF32swap::WC2MB(char *buf, const wchar_t *psz, size_t n) const
712{
713 size_t len=0;
714
715 while (*psz && (!buf || len < n))
716 {
717 char cc[4];
718
b5153fd8
VZ
719 // cast is ok for WC_UTF16
720 size_t pa=decode_utf16((const wxUint16 *)psz, *(wxUint32*)cc);
c91830cb
VZ
721 if (pa == (size_t)-1)
722 return pa;
723
724 if (buf)
725 {
726 *buf++ = cc[3];
727 *buf++ = cc[2];
728 *buf++ = cc[1];
729 *buf++ = cc[0];
730 }
731 len += sizeof(wxUint32);
732 psz += pa;
733 }
b5153fd8
VZ
734
735 if (buf && len<=n-sizeof(wxUint32))
736 *(wxUint32*)buf=0;
c91830cb
VZ
737
738 return len;
739}
740
741#else // WC_UTF16
742
743
744// copy 32bit MB to 32bit String
745size_t wxMBConvUTF32straight::MB2WC(wchar_t *buf, const char *psz, size_t n) const
746{
747 size_t len=0;
748
749 while (*(wxUint32*)psz && (!buf || len < n))
750 {
751 if (buf)
752 *buf++ = *(wxUint32*)psz;
753 len++;
754 psz += sizeof(wxUint32);
755 }
b5153fd8
VZ
756
757 if (buf && len<n)
758 *buf=0;
c91830cb
VZ
759
760 return len;
761}
762
763
764// copy 32bit String to 32bit MB
765size_t wxMBConvUTF32straight::WC2MB(char *buf, const wchar_t *psz, size_t n) const
766{
767 size_t len=0;
768
769 while (*psz && (!buf || len < n))
770 {
771 if (buf)
772 {
773 *(wxUint32*)buf = *psz;
774 buf += sizeof(wxUint32);
775 }
776
777 len += sizeof(wxUint32);
778 psz++;
779 }
780
b5153fd8
VZ
781 if (buf && len<=n-sizeof(wxUint32))
782 *(wxUint32*)buf=0;
c91830cb
VZ
783
784 return len;
785}
786
787
788// swap 32bit MB to 32bit String
789size_t wxMBConvUTF32swap::MB2WC(wchar_t *buf, const char *psz, size_t n) const
790{
791 size_t len=0;
792
793 while (*(wxUint32*)psz && (!buf || len < n))
794 {
795 if (buf)
796 {
797 ((char *)buf)[0] = psz[3];
798 ((char *)buf)[1] = psz[2];
799 ((char *)buf)[2] = psz[1];
800 ((char *)buf)[3] = psz[0];
801 buf++;
802 }
803 len++;
804 psz += sizeof(wxUint32);
805 }
b5153fd8
VZ
806
807 if (buf && len<n)
808 *buf=0;
c91830cb
VZ
809
810 return len;
811}
812
813
814// swap 32bit String to 32bit MB
815size_t wxMBConvUTF32swap::WC2MB(char *buf, const wchar_t *psz, size_t n) const
816{
817 size_t len=0;
818
819 while (*psz && (!buf || len < n))
820 {
821 if (buf)
822 {
823 *buf++ = ((char *)psz)[3];
824 *buf++ = ((char *)psz)[2];
825 *buf++ = ((char *)psz)[1];
826 *buf++ = ((char *)psz)[0];
827 }
828 len += sizeof(wxUint32);
829 psz++;
830 }
b5153fd8
VZ
831
832 if (buf && len<=n-sizeof(wxUint32))
833 *(wxUint32*)buf=0;
c91830cb
VZ
834
835 return len;
836}
837
838
839#endif // WC_UTF16
840
841
36acb880
VZ
842// ============================================================================
843// The classes doing conversion using the iconv_xxx() functions
844// ============================================================================
3caec1bb 845
b040e242 846#ifdef HAVE_ICONV
3a0d76bc 847
3caec1bb
VS
848// VS: glibc 2.1.3 is broken in that iconv() conversion to/from UCS4 fails with E2BIG
849// if output buffer is _exactly_ as big as needed. Such case is (unless there's
850// yet another bug in glibc) the only case when iconv() returns with (size_t)-1
851// (which means error) and says there are 0 bytes left in the input buffer --
852// when _real_ error occurs, bytes-left-in-input buffer is non-zero. Hence,
853// this alternative test for iconv() failure.
854// [This bug does not appear in glibc 2.2.]
855#if defined(__GLIBC__) && __GLIBC__ == 2 && __GLIBC_MINOR__ <= 1
856#define ICONV_FAILED(cres, bufLeft) ((cres == (size_t)-1) && \
857 (errno != E2BIG || bufLeft != 0))
858#else
859#define ICONV_FAILED(cres, bufLeft) (cres == (size_t)-1)
860#endif
861
ab217dba 862#define ICONV_CHAR_CAST(x) ((ICONV_CONST char **)(x))
36acb880
VZ
863
864// ----------------------------------------------------------------------------
e95354ec 865// wxMBConv_iconv: encapsulates an iconv character set
36acb880
VZ
866// ----------------------------------------------------------------------------
867
e95354ec 868class wxMBConv_iconv : public wxMBConv
1cd52418
OK
869{
870public:
e95354ec
VZ
871 wxMBConv_iconv(const wxChar *name);
872 virtual ~wxMBConv_iconv();
36acb880 873
bde4baac
VZ
874 virtual size_t MB2WC(wchar_t *buf, const char *psz, size_t n) const;
875 virtual size_t WC2MB(char *buf, const wchar_t *psz, size_t n) const;
36acb880 876
e95354ec 877 bool IsOk() const
36acb880
VZ
878 { return (m2w != (iconv_t)-1) && (w2m != (iconv_t)-1); }
879
880protected:
881 // the iconv handlers used to translate from multibyte to wide char and in
882 // the other direction
883 iconv_t m2w,
884 w2m;
885
886private:
e95354ec 887 // the name (for iconv_open()) of a wide char charset -- if none is
36acb880
VZ
888 // available on this machine, it will remain NULL
889 static const char *ms_wcCharsetName;
890
891 // true if the wide char encoding we use (i.e. ms_wcCharsetName) has
892 // different endian-ness than the native one
405d8f46 893 static bool ms_wcNeedsSwap;
36acb880
VZ
894};
895
e95354ec
VZ
896const char *wxMBConv_iconv::ms_wcCharsetName = NULL;
897bool wxMBConv_iconv::ms_wcNeedsSwap = false;
36acb880 898
e95354ec 899wxMBConv_iconv::wxMBConv_iconv(const wxChar *name)
36acb880 900{
04c79127
RR
901 // Do it the hard way
902 char cname[100];
903 for (size_t i = 0; i < wxStrlen(name)+1; i++)
904 cname[i] = (char) name[i];
905
36acb880
VZ
906 // check for charset that represents wchar_t:
907 if (ms_wcCharsetName == NULL)
f1339c56 908 {
e95354ec 909 ms_wcNeedsSwap = false;
dccce9ea 910
36acb880
VZ
911 // try charset with explicit bytesex info (e.g. "UCS-4LE"):
912 ms_wcCharsetName = WC_NAME_BEST;
04c79127 913 m2w = iconv_open(ms_wcCharsetName, cname);
3a0d76bc 914
36acb880
VZ
915 if (m2w == (iconv_t)-1)
916 {
917 // try charset w/o bytesex info (e.g. "UCS4")
918 // and check for bytesex ourselves:
919 ms_wcCharsetName = WC_NAME;
04c79127 920 m2w = iconv_open(ms_wcCharsetName, cname);
36acb880
VZ
921
922 // last bet, try if it knows WCHAR_T pseudo-charset
3a0d76bc
VS
923 if (m2w == (iconv_t)-1)
924 {
36acb880 925 ms_wcCharsetName = "WCHAR_T";
04c79127 926 m2w = iconv_open(ms_wcCharsetName, cname);
36acb880 927 }
3a0d76bc 928
36acb880
VZ
929 if (m2w != (iconv_t)-1)
930 {
931 char buf[2], *bufPtr;
932 wchar_t wbuf[2], *wbufPtr;
933 size_t insz, outsz;
934 size_t res;
935
936 buf[0] = 'A';
937 buf[1] = 0;
938 wbuf[0] = 0;
939 insz = 2;
940 outsz = SIZEOF_WCHAR_T * 2;
941 wbufPtr = wbuf;
942 bufPtr = buf;
943
944 res = iconv(m2w, ICONV_CHAR_CAST(&bufPtr), &insz,
945 (char**)&wbufPtr, &outsz);
946
947 if (ICONV_FAILED(res, insz))
3a0d76bc 948 {
36acb880
VZ
949 ms_wcCharsetName = NULL;
950 wxLogLastError(wxT("iconv"));
2b5f62a0 951 wxLogError(_("Conversion to charset '%s' doesn't work."), name);
3a0d76bc
VS
952 }
953 else
954 {
36acb880 955 ms_wcNeedsSwap = wbuf[0] != (wchar_t)buf[0];
3a0d76bc
VS
956 }
957 }
36acb880
VZ
958 else
959 {
960 ms_wcCharsetName = NULL;
373658eb 961
77ffb593 962 // VS: we must not output an error here, since wxWidgets will safely
957686c8
VS
963 // fall back to using wxEncodingConverter.
964 wxLogTrace(wxT("strconv"), wxT("Impossible to convert to/from charset '%s' with iconv, falling back to wxEncodingConverter."), name);
965 //wxLogError(
36acb880 966 }
3a0d76bc 967 }
36acb880 968 wxLogTrace(wxT("strconv"), wxT("wchar_t charset is '%s', needs swap: %i"), ms_wcCharsetName, ms_wcNeedsSwap);
3a0d76bc 969 }
36acb880 970 else // we already have ms_wcCharsetName
3caec1bb 971 {
04c79127 972 m2w = iconv_open(ms_wcCharsetName, cname);
f1339c56 973 }
dccce9ea 974
36acb880
VZ
975 // NB: don't ever pass NULL to iconv_open(), it may crash!
976 if ( ms_wcCharsetName )
f1339c56 977 {
04c79127 978 w2m = iconv_open( cname, ms_wcCharsetName);
36acb880 979 }
405d8f46
VZ
980 else
981 {
982 w2m = (iconv_t)-1;
983 }
36acb880 984}
3caec1bb 985
e95354ec 986wxMBConv_iconv::~wxMBConv_iconv()
36acb880
VZ
987{
988 if ( m2w != (iconv_t)-1 )
989 iconv_close(m2w);
990 if ( w2m != (iconv_t)-1 )
991 iconv_close(w2m);
992}
3a0d76bc 993
bde4baac 994size_t wxMBConv_iconv::MB2WC(wchar_t *buf, const char *psz, size_t n) const
36acb880
VZ
995{
996 size_t inbuf = strlen(psz);
997 size_t outbuf = n * SIZEOF_WCHAR_T;
998 size_t res, cres;
999 // VS: Use these instead of psz, buf because iconv() modifies its arguments:
1000 wchar_t *bufPtr = buf;
1001 const char *pszPtr = psz;
1002
1003 if (buf)
1004 {
1005 // have destination buffer, convert there
1006 cres = iconv(m2w,
1007 ICONV_CHAR_CAST(&pszPtr), &inbuf,
1008 (char**)&bufPtr, &outbuf);
1009 res = n - (outbuf / SIZEOF_WCHAR_T);
dccce9ea 1010
36acb880 1011 if (ms_wcNeedsSwap)
3a0d76bc 1012 {
36acb880
VZ
1013 // convert to native endianness
1014 WC_BSWAP(buf /* _not_ bufPtr */, res)
3a0d76bc 1015 }
adb45366 1016
49dd9820
VS
1017 // NB: iconv was given only strlen(psz) characters on input, and so
1018 // it couldn't convert the trailing zero. Let's do it ourselves
1019 // if there's some room left for it in the output buffer.
1020 if (res < n)
1021 buf[res] = 0;
36acb880
VZ
1022 }
1023 else
1024 {
1025 // no destination buffer... convert using temp buffer
1026 // to calculate destination buffer requirement
1027 wchar_t tbuf[8];
1028 res = 0;
1029 do {
1030 bufPtr = tbuf;
1031 outbuf = 8*SIZEOF_WCHAR_T;
1032
1033 cres = iconv(m2w,
1034 ICONV_CHAR_CAST(&pszPtr), &inbuf,
1035 (char**)&bufPtr, &outbuf );
1036
1037 res += 8-(outbuf/SIZEOF_WCHAR_T);
1038 } while ((cres==(size_t)-1) && (errno==E2BIG));
f1339c56 1039 }
dccce9ea 1040
36acb880 1041 if (ICONV_FAILED(cres, inbuf))
f1339c56 1042 {
36acb880
VZ
1043 //VS: it is ok if iconv fails, hence trace only
1044 wxLogTrace(wxT("strconv"), wxT("iconv failed: %s"), wxSysErrorMsg(wxSysErrorCode()));
1045 return (size_t)-1;
1046 }
1047
1048 return res;
1049}
1050
bde4baac 1051size_t wxMBConv_iconv::WC2MB(char *buf, const wchar_t *psz, size_t n) const
36acb880 1052{
f8d791e0 1053 size_t inbuf = wxWcslen(psz) * SIZEOF_WCHAR_T;
36acb880
VZ
1054 size_t outbuf = n;
1055 size_t res, cres;
3a0d76bc 1056
36acb880 1057 wchar_t *tmpbuf = 0;
3caec1bb 1058
36acb880
VZ
1059 if (ms_wcNeedsSwap)
1060 {
1061 // need to copy to temp buffer to switch endianness
1062 // this absolutely doesn't rock!
1063 // (no, doing WC_BSWAP twice on the original buffer won't help, as it
1064 // could be in read-only memory, or be accessed in some other thread)
1065 tmpbuf=(wchar_t*)malloc((inbuf+1)*SIZEOF_WCHAR_T);
1066 memcpy(tmpbuf,psz,(inbuf+1)*SIZEOF_WCHAR_T);
1067 WC_BSWAP(tmpbuf, inbuf)
1068 psz=tmpbuf;
1069 }
3a0d76bc 1070
36acb880
VZ
1071 if (buf)
1072 {
1073 // have destination buffer, convert there
1074 cres = iconv( w2m, ICONV_CHAR_CAST(&psz), &inbuf, &buf, &outbuf );
3a0d76bc 1075
36acb880 1076 res = n-outbuf;
adb45366 1077
49dd9820
VS
1078 // NB: iconv was given only wcslen(psz) characters on input, and so
1079 // it couldn't convert the trailing zero. Let's do it ourselves
1080 // if there's some room left for it in the output buffer.
1081 if (res < n)
1082 buf[0] = 0;
36acb880
VZ
1083 }
1084 else
1085 {
1086 // no destination buffer... convert using temp buffer
1087 // to calculate destination buffer requirement
1088 char tbuf[16];
1089 res = 0;
1090 do {
1091 buf = tbuf; outbuf = 16;
1092
1093 cres = iconv( w2m, ICONV_CHAR_CAST(&psz), &inbuf, &buf, &outbuf );
dccce9ea 1094
36acb880
VZ
1095 res += 16 - outbuf;
1096 } while ((cres==(size_t)-1) && (errno==E2BIG));
f1339c56 1097 }
dccce9ea 1098
36acb880
VZ
1099 if (ms_wcNeedsSwap)
1100 {
1101 free(tmpbuf);
1102 }
dccce9ea 1103
36acb880
VZ
1104 if (ICONV_FAILED(cres, inbuf))
1105 {
1106 //VS: it is ok if iconv fails, hence trace only
1107 wxLogTrace(wxT("strconv"), wxT("iconv failed: %s"), wxSysErrorMsg(wxSysErrorCode()));
1108 return (size_t)-1;
1109 }
1110
1111 return res;
1112}
1113
b040e242 1114#endif // HAVE_ICONV
36acb880 1115
e95354ec 1116
36acb880
VZ
1117// ============================================================================
1118// Win32 conversion classes
1119// ============================================================================
1cd52418 1120
e95354ec 1121#ifdef wxHAVE_WIN32_MB2WC
373658eb 1122
8b04d4c4 1123// from utils.cpp
d775fa82 1124#if wxUSE_FONTMAP
8b04d4c4
VZ
1125extern WXDLLIMPEXP_BASE long wxCharsetToCodepage(const wxChar *charset);
1126extern WXDLLIMPEXP_BASE long wxEncodingToCodepage(wxFontEncoding encoding);
7608a683 1127#endif
373658eb 1128
e95354ec 1129class wxMBConv_win32 : public wxMBConv
1cd52418
OK
1130{
1131public:
bde4baac
VZ
1132 wxMBConv_win32()
1133 {
1134 m_CodePage = CP_ACP;
1135 }
1136
7608a683 1137#if wxUSE_FONTMAP
e95354ec 1138 wxMBConv_win32(const wxChar* name)
bde4baac
VZ
1139 {
1140 m_CodePage = wxCharsetToCodepage(name);
1141 }
dccce9ea 1142
e95354ec 1143 wxMBConv_win32(wxFontEncoding encoding)
bde4baac
VZ
1144 {
1145 m_CodePage = wxEncodingToCodepage(encoding);
1146 }
7608a683 1147#endif
8b04d4c4 1148
bde4baac 1149 size_t MB2WC(wchar_t *buf, const char *psz, size_t n) const
f1339c56 1150 {
02272c9c
VZ
1151 // note that we have to use MB_ERR_INVALID_CHARS flag as it without it
1152 // the behaviour is not compatible with the Unix version (using iconv)
1153 // and break the library itself, e.g. wxTextInputStream::NextChar()
1154 // wouldn't work if reading an incomplete MB char didn't result in an
1155 // error
2b5f62a0
VZ
1156 const size_t len = ::MultiByteToWideChar
1157 (
1158 m_CodePage, // code page
02272c9c 1159 MB_ERR_INVALID_CHARS, // flags: fall on error
2b5f62a0
VZ
1160 psz, // input string
1161 -1, // its length (NUL-terminated)
b4da152e 1162 buf, // output string
2b5f62a0
VZ
1163 buf ? n : 0 // size of output buffer
1164 );
1165
03a991bc
VZ
1166 // note that it returns count of written chars for buf != NULL and size
1167 // of the needed buffer for buf == NULL so in either case the length of
1168 // the string (which never includes the terminating NUL) is one less
1169 return len ? len - 1 : (size_t)-1;
f1339c56 1170 }
dccce9ea 1171
13dd924a 1172 size_t WC2MB(char *buf, const wchar_t *pwz, size_t n) const
f1339c56 1173 {
13dd924a
VZ
1174 /*
1175 we have a problem here: by default, WideCharToMultiByte() may
1176 replace characters unrepresentable in the target code page with bad
1177 quality approximations such as turning "1/2" symbol (U+00BD) into
1178 "1" for the code pages which don't have it and we, obviously, want
1179 to avoid this at any price
d775fa82 1180
13dd924a
VZ
1181 the trouble is that this function does it _silently_, i.e. it won't
1182 even tell us whether it did or not... Win98/2000 and higher provide
1183 WC_NO_BEST_FIT_CHARS but it doesn't work for the older systems and
1184 we have to resort to a round trip, i.e. check that converting back
1185 results in the same string -- this is, of course, expensive but
1186 otherwise we simply can't be sure to not garble the data.
1187 */
1188
1189 // determine if we can rely on WC_NO_BEST_FIT_CHARS: according to MSDN
1190 // it doesn't work with CJK encodings (which we test for rather roughly
1191 // here...) nor with UTF-7/8 nor, of course, with Windows versions not
1192 // supporting it
1193 BOOL usedDef wxDUMMY_INITIALIZE(false),
1194 *pUsedDef;
1195 int flags;
1196 if ( CanUseNoBestFit() && m_CodePage < 50000 )
1197 {
1198 // it's our lucky day
1199 flags = WC_NO_BEST_FIT_CHARS;
1200 pUsedDef = &usedDef;
1201 }
1202 else // old system or unsupported encoding
1203 {
1204 flags = 0;
1205 pUsedDef = NULL;
1206 }
1207
2b5f62a0
VZ
1208 const size_t len = ::WideCharToMultiByte
1209 (
1210 m_CodePage, // code page
13dd924a
VZ
1211 flags, // either none or no best fit
1212 pwz, // input string
2b5f62a0
VZ
1213 -1, // it is (wide) NUL-terminated
1214 buf, // output buffer
1215 buf ? n : 0, // and its size
1216 NULL, // default "replacement" char
13dd924a 1217 pUsedDef // [out] was it used?
2b5f62a0
VZ
1218 );
1219
13dd924a
VZ
1220 if ( !len )
1221 {
1222 // function totally failed
1223 return (size_t)-1;
1224 }
1225
1226 // if we were really converting, check if we succeeded
1227 if ( buf )
1228 {
1229 if ( flags )
1230 {
1231 // check if the conversion failed, i.e. if any replacements
1232 // were done
1233 if ( usedDef )
1234 return (size_t)-1;
1235 }
1236 else // we must resort to double tripping...
1237 {
1238 wxWCharBuffer wcBuf(n);
1239 if ( MB2WC(wcBuf.data(), buf, n) == (size_t)-1 ||
1240 wcscmp(wcBuf, pwz) != 0 )
1241 {
1242 // we didn't obtain the same thing we started from, hence
1243 // the conversion was lossy and we consider that it failed
1244 return (size_t)-1;
1245 }
1246 }
1247 }
1248
03a991bc 1249 // see the comment above for the reason of "len - 1"
13dd924a 1250 return len - 1;
f1339c56 1251 }
dccce9ea 1252
13dd924a
VZ
1253 bool IsOk() const { return m_CodePage != -1; }
1254
1255private:
1256 static bool CanUseNoBestFit()
1257 {
1258 static int s_isWin98Or2k = -1;
1259
1260 if ( s_isWin98Or2k == -1 )
1261 {
1262 int verMaj, verMin;
1263 switch ( wxGetOsVersion(&verMaj, &verMin) )
1264 {
1265 case wxWIN95:
1266 s_isWin98Or2k = verMaj >= 4 && verMin >= 10;
1267 break;
1268
1269 case wxWINDOWS_NT:
1270 s_isWin98Or2k = verMaj >= 5;
1271 break;
1272
1273 default:
1274 // unknown, be conseravtive by default
1275 s_isWin98Or2k = 0;
1276 }
1277
1278 wxASSERT_MSG( s_isWin98Or2k != -1, _T("should be set above") );
1279 }
1280
1281 return s_isWin98Or2k == 1;
1282 }
f1339c56 1283
b1d66b54 1284 long m_CodePage;
1cd52418 1285};
e95354ec
VZ
1286
1287#endif // wxHAVE_WIN32_MB2WC
1288
335d31e0
SC
1289// ============================================================================
1290// Mac conversion classes
1291// ============================================================================
1292
1293#if defined(__WXMAC__) && defined(TARGET_CARBON)
1294
1295class wxMBConv_mac : public wxMBConv
1296{
1297public:
1298 wxMBConv_mac()
1299 {
1300 Init(CFStringGetSystemEncoding()) ;
1301 }
1302
1303 wxMBConv_mac(const wxChar* name)
1304 {
d775fa82 1305 Init( wxMacGetSystemEncFromFontEnc(wxFontMapper::Get()->CharsetToEncoding(name, false) ) ) ;
335d31e0
SC
1306 }
1307
1308 wxMBConv_mac(wxFontEncoding encoding)
1309 {
d775fa82
WS
1310 Init( wxMacGetSystemEncFromFontEnc(encoding) );
1311 }
1312
1313 ~wxMBConv_mac()
1314 {
1315 OSStatus status = noErr ;
1316 status = TECDisposeConverter(m_MB2WC_converter);
1317 status = TECDisposeConverter(m_WC2MB_converter);
1318 }
1319
1320
1321 void Init( TextEncodingBase encoding)
1322 {
1323 OSStatus status = noErr ;
1324 m_char_encoding = encoding ;
1325 m_unicode_encoding = CreateTextEncoding(kTextEncodingUnicodeDefault,0,kUnicode16BitFormat) ;
1326
1327 status = TECCreateConverter(&m_MB2WC_converter,
1328 m_char_encoding,
1329 m_unicode_encoding);
1330 status = TECCreateConverter(&m_WC2MB_converter,
1331 m_unicode_encoding,
1332 m_char_encoding);
1333 }
1334
335d31e0
SC
1335 size_t MB2WC(wchar_t *buf, const char *psz, size_t n) const
1336 {
d775fa82
WS
1337 OSStatus status = noErr ;
1338 ByteCount byteOutLen ;
1339 ByteCount byteInLen = strlen(psz) ;
1340 wchar_t *tbuf = NULL ;
1341 UniChar* ubuf = NULL ;
1342 size_t res = 0 ;
1343
1344 if (buf == NULL)
1345 {
1346 n = byteInLen ;
1347 tbuf = (wchar_t*) malloc( n * SIZEOF_WCHAR_T) ;
1348 }
1349 ByteCount byteBufferLen = n * sizeof( UniChar ) ;
f3a355ce 1350#if SIZEOF_WCHAR_T == 4
d775fa82 1351 ubuf = (UniChar*) malloc( byteBufferLen + 2 ) ;
f3a355ce 1352#else
d775fa82 1353 ubuf = (UniChar*) (buf ? buf : tbuf) ;
f3a355ce 1354#endif
d775fa82
WS
1355 status = TECConvertText(m_MB2WC_converter, (ConstTextPtr) psz , byteInLen, &byteInLen,
1356 (TextPtr) ubuf , byteBufferLen, &byteOutLen);
f3a355ce 1357#if SIZEOF_WCHAR_T == 4
8471ea90
SC
1358 // we have to terminate here, because n might be larger for the trailing zero, and if UniChar
1359 // is not properly terminated we get random characters at the end
1360 ubuf[byteOutLen / sizeof( UniChar ) ] = 0 ;
d775fa82
WS
1361 wxMBConvUTF16BE converter ;
1362 res = converter.MB2WC( (buf ? buf : tbuf) , (const char*)ubuf , n ) ;
1363 free( ubuf ) ;
f3a355ce 1364#else
d775fa82 1365 res = byteOutLen / sizeof( UniChar ) ;
f3a355ce 1366#endif
d775fa82
WS
1367 if ( buf == NULL )
1368 free(tbuf) ;
335d31e0 1369
335d31e0
SC
1370 if ( buf && res < n)
1371 buf[res] = 0;
1372
d775fa82 1373 return res ;
335d31e0
SC
1374 }
1375
1376 size_t WC2MB(char *buf, const wchar_t *psz, size_t n) const
d775fa82
WS
1377 {
1378 OSStatus status = noErr ;
1379 ByteCount byteOutLen ;
1380 ByteCount byteInLen = wxWcslen(psz) * SIZEOF_WCHAR_T ;
1381
1382 char *tbuf = NULL ;
1383
1384 if (buf == NULL)
1385 {
1386 // worst case
1387 n = byteInLen * 2 ;
1388 tbuf = (char*) malloc( n ) ;
1389 }
1390
1391 ByteCount byteBufferLen = n ;
1392 UniChar* ubuf = NULL ;
f3a355ce 1393#if SIZEOF_WCHAR_T == 4
d775fa82
WS
1394 wxMBConvUTF16BE converter ;
1395 size_t unicharlen = converter.WC2MB( NULL , psz , 0 ) ;
1396 byteInLen = unicharlen ;
1397 ubuf = (UniChar*) malloc( byteInLen + 2 ) ;
1398 converter.WC2MB( (char*) ubuf , psz, unicharlen + 2 ) ;
f3a355ce 1399#else
d775fa82 1400 ubuf = (UniChar*) psz ;
f3a355ce 1401#endif
d775fa82
WS
1402 status = TECConvertText(m_WC2MB_converter, (ConstTextPtr) ubuf , byteInLen, &byteInLen,
1403 (TextPtr) (buf ? buf : tbuf) , byteBufferLen, &byteOutLen);
f3a355ce 1404#if SIZEOF_WCHAR_T == 4
d775fa82 1405 free( ubuf ) ;
f3a355ce 1406#endif
d775fa82
WS
1407 if ( buf == NULL )
1408 free(tbuf) ;
335d31e0 1409
d775fa82 1410 size_t res = byteOutLen ;
335d31e0
SC
1411 if ( buf && res < n)
1412 buf[res] = 0;
1413
d775fa82 1414 return res ;
335d31e0
SC
1415 }
1416
1417 bool IsOk() const
1418 { return m_MB2WC_converter != NULL && m_WC2MB_converter != NULL ; }
1419
1420private:
d775fa82
WS
1421 TECObjectRef m_MB2WC_converter ;
1422 TECObjectRef m_WC2MB_converter ;
1423
1424 TextEncodingBase m_char_encoding ;
1425 TextEncodingBase m_unicode_encoding ;
335d31e0
SC
1426};
1427
1428#endif // defined(__WXMAC__) && defined(TARGET_CARBON)
1e6feb95 1429
36acb880
VZ
1430// ============================================================================
1431// wxEncodingConverter based conversion classes
1432// ============================================================================
1433
1e6feb95 1434#if wxUSE_FONTMAP
1cd52418 1435
e95354ec 1436class wxMBConv_wxwin : public wxMBConv
1cd52418 1437{
8b04d4c4
VZ
1438private:
1439 void Init()
1440 {
1441 m_ok = m2w.Init(m_enc, wxFONTENCODING_UNICODE) &&
1442 w2m.Init(wxFONTENCODING_UNICODE, m_enc);
1443 }
1444
6001e347 1445public:
f1339c56
RR
1446 // temporarily just use wxEncodingConverter stuff,
1447 // so that it works while a better implementation is built
e95354ec 1448 wxMBConv_wxwin(const wxChar* name)
f1339c56
RR
1449 {
1450 if (name)
e95354ec 1451 m_enc = wxFontMapper::Get()->CharsetToEncoding(name, false);
8b04d4c4
VZ
1452 else
1453 m_enc = wxFONTENCODING_SYSTEM;
cafbf6fb 1454
8b04d4c4
VZ
1455 Init();
1456 }
1457
e95354ec 1458 wxMBConv_wxwin(wxFontEncoding enc)
8b04d4c4
VZ
1459 {
1460 m_enc = enc;
1461
1462 Init();
f1339c56 1463 }
dccce9ea 1464
bde4baac 1465 size_t MB2WC(wchar_t *buf, const char *psz, size_t WXUNUSED(n)) const
f1339c56
RR
1466 {
1467 size_t inbuf = strlen(psz);
dccce9ea 1468 if (buf)
4def3b35 1469 m2w.Convert(psz,buf);
f1339c56
RR
1470 return inbuf;
1471 }
dccce9ea 1472
bde4baac 1473 size_t WC2MB(char *buf, const wchar_t *psz, size_t WXUNUSED(n)) const
f1339c56 1474 {
f8d791e0 1475 const size_t inbuf = wxWcslen(psz);
f1339c56
RR
1476 if (buf)
1477 w2m.Convert(psz,buf);
dccce9ea 1478
f1339c56
RR
1479 return inbuf;
1480 }
dccce9ea 1481
e95354ec 1482 bool IsOk() const { return m_ok; }
f1339c56
RR
1483
1484public:
8b04d4c4 1485 wxFontEncoding m_enc;
f1339c56 1486 wxEncodingConverter m2w, w2m;
cafbf6fb
VZ
1487
1488 // were we initialized successfully?
1489 bool m_ok;
fc7a2a60 1490
e95354ec 1491 DECLARE_NO_COPY_CLASS(wxMBConv_wxwin)
f6bcfd97 1492};
6001e347 1493
1e6feb95
VZ
1494#endif // wxUSE_FONTMAP
1495
36acb880
VZ
1496// ============================================================================
1497// wxCSConv implementation
1498// ============================================================================
1499
8b04d4c4 1500void wxCSConv::Init()
6001e347 1501{
e95354ec
VZ
1502 m_name = NULL;
1503 m_convReal = NULL;
1504 m_deferred = true;
1505}
1506
8b04d4c4
VZ
1507wxCSConv::wxCSConv(const wxChar *charset)
1508{
1509 Init();
82713003 1510
e95354ec
VZ
1511 if ( charset )
1512 {
e95354ec
VZ
1513 SetName(charset);
1514 }
bda3d86a
VZ
1515
1516 m_encoding = wxFONTENCODING_SYSTEM;
6001e347
RR
1517}
1518
8b04d4c4
VZ
1519wxCSConv::wxCSConv(wxFontEncoding encoding)
1520{
bda3d86a 1521 if ( encoding == wxFONTENCODING_MAX || encoding == wxFONTENCODING_DEFAULT )
e95354ec
VZ
1522 {
1523 wxFAIL_MSG( _T("invalid encoding value in wxCSConv ctor") );
1524
1525 encoding = wxFONTENCODING_SYSTEM;
1526 }
1527
8b04d4c4
VZ
1528 Init();
1529
bda3d86a 1530 m_encoding = encoding;
8b04d4c4
VZ
1531}
1532
6001e347
RR
1533wxCSConv::~wxCSConv()
1534{
65e50848
JS
1535 Clear();
1536}
1537
54380f29 1538wxCSConv::wxCSConv(const wxCSConv& conv)
8b04d4c4 1539 : wxMBConv()
54380f29 1540{
8b04d4c4
VZ
1541 Init();
1542
54380f29 1543 SetName(conv.m_name);
8b04d4c4 1544 m_encoding = conv.m_encoding;
54380f29
GD
1545}
1546
1547wxCSConv& wxCSConv::operator=(const wxCSConv& conv)
1548{
1549 Clear();
8b04d4c4 1550
54380f29 1551 SetName(conv.m_name);
8b04d4c4
VZ
1552 m_encoding = conv.m_encoding;
1553
54380f29
GD
1554 return *this;
1555}
1556
65e50848
JS
1557void wxCSConv::Clear()
1558{
8b04d4c4 1559 free(m_name);
e95354ec 1560 delete m_convReal;
8b04d4c4 1561
65e50848 1562 m_name = NULL;
e95354ec 1563 m_convReal = NULL;
6001e347
RR
1564}
1565
1566void wxCSConv::SetName(const wxChar *charset)
1567{
f1339c56
RR
1568 if (charset)
1569 {
1570 m_name = wxStrdup(charset);
e95354ec 1571 m_deferred = true;
f1339c56 1572 }
6001e347
RR
1573}
1574
e95354ec
VZ
1575wxMBConv *wxCSConv::DoCreate() const
1576{
c547282d
VZ
1577 // check for the special case of ASCII or ISO8859-1 charset: as we have
1578 // special knowledge of it anyhow, we don't need to create a special
1579 // conversion object
1580 if ( m_encoding == wxFONTENCODING_ISO8859_1 )
f1339c56 1581 {
e95354ec
VZ
1582 // don't convert at all
1583 return NULL;
1584 }
dccce9ea 1585
e95354ec
VZ
1586 // we trust OS to do conversion better than we can so try external
1587 // conversion methods first
1588 //
1589 // the full order is:
1590 // 1. OS conversion (iconv() under Unix or Win32 API)
1591 // 2. hard coded conversions for UTF
1592 // 3. wxEncodingConverter as fall back
1593
1594 // step (1)
1595#ifdef HAVE_ICONV
c547282d 1596#if !wxUSE_FONTMAP
e95354ec 1597 if ( m_name )
c547282d 1598#endif // !wxUSE_FONTMAP
e95354ec 1599 {
c547282d
VZ
1600 wxString name(m_name);
1601
1602#if wxUSE_FONTMAP
1603 if ( name.empty() )
1604 name = wxFontMapper::Get()->GetEncodingName(m_encoding);
1605#endif // wxUSE_FONTMAP
1606
1607 wxMBConv_iconv *conv = new wxMBConv_iconv(name);
e95354ec
VZ
1608 if ( conv->IsOk() )
1609 return conv;
1610
1611 delete conv;
1612 }
1613#endif // HAVE_ICONV
1614
1615#ifdef wxHAVE_WIN32_MB2WC
1616 {
7608a683 1617#if wxUSE_FONTMAP
e95354ec
VZ
1618 wxMBConv_win32 *conv = m_name ? new wxMBConv_win32(m_name)
1619 : new wxMBConv_win32(m_encoding);
1620 if ( conv->IsOk() )
1621 return conv;
1622
1623 delete conv;
7608a683
WS
1624#else
1625 return NULL;
1626#endif
e95354ec
VZ
1627 }
1628#endif // wxHAVE_WIN32_MB2WC
d775fa82
WS
1629#if defined(__WXMAC__)
1630 {
1631 if ( m_name || ( m_encoding < wxFONTENCODING_UTF16BE ) )
1632 {
1633
1634 wxMBConv_mac *conv = m_name ? new wxMBConv_mac(m_name)
1635 : new wxMBConv_mac(m_encoding);
1636 if ( conv->IsOk() )
1637 return conv;
1638
1639 delete conv;
1640 }
335d31e0
SC
1641 }
1642#endif
e95354ec
VZ
1643 // step (2)
1644 wxFontEncoding enc = m_encoding;
1645#if wxUSE_FONTMAP
c547282d
VZ
1646 if ( enc == wxFONTENCODING_SYSTEM && m_name )
1647 {
1648 // use "false" to suppress interactive dialogs -- we can be called from
1649 // anywhere and popping up a dialog from here is the last thing we want to
1650 // do
1651 enc = wxFontMapper::Get()->CharsetToEncoding(m_name, false);
1652 }
e95354ec
VZ
1653#endif // wxUSE_FONTMAP
1654
1655 switch ( enc )
1656 {
1657 case wxFONTENCODING_UTF7:
1658 return new wxMBConvUTF7;
1659
1660 case wxFONTENCODING_UTF8:
1661 return new wxMBConvUTF8;
1662
e95354ec
VZ
1663 case wxFONTENCODING_UTF16BE:
1664 return new wxMBConvUTF16BE;
1665
1666 case wxFONTENCODING_UTF16LE:
1667 return new wxMBConvUTF16LE;
1668
e95354ec
VZ
1669 case wxFONTENCODING_UTF32BE:
1670 return new wxMBConvUTF32BE;
1671
1672 case wxFONTENCODING_UTF32LE:
1673 return new wxMBConvUTF32LE;
1674
1675 default:
1676 // nothing to do but put here to suppress gcc warnings
1677 ;
1678 }
1679
1680 // step (3)
1681#if wxUSE_FONTMAP
1682 {
1683 wxMBConv_wxwin *conv = m_name ? new wxMBConv_wxwin(m_name)
1684 : new wxMBConv_wxwin(m_encoding);
1685 if ( conv->IsOk() )
1686 return conv;
1687
1688 delete conv;
1689 }
1690#endif // wxUSE_FONTMAP
1691
a58d4f4d
VS
1692 // NB: This is a hack to prevent deadlock. What could otherwise happen
1693 // in Unicode build: wxConvLocal creation ends up being here
1694 // because of some failure and logs the error. But wxLog will try to
1695 // attach timestamp, for which it will need wxConvLocal (to convert
1696 // time to char* and then wchar_t*), but that fails, tries to log
1697 // error, but wxLog has a (already locked) critical section that
1698 // guards static buffer.
1699 static bool alreadyLoggingError = false;
1700 if (!alreadyLoggingError)
1701 {
1702 alreadyLoggingError = true;
1703 wxLogError(_("Cannot convert from the charset '%s'!"),
1704 m_name ? m_name
e95354ec
VZ
1705 :
1706#if wxUSE_FONTMAP
1707 wxFontMapper::GetEncodingDescription(m_encoding).c_str()
1708#else // !wxUSE_FONTMAP
1709 wxString::Format(_("encoding %s"), m_encoding).c_str()
1710#endif // wxUSE_FONTMAP/!wxUSE_FONTMAP
1711 );
a58d4f4d
VS
1712 alreadyLoggingError = false;
1713 }
e95354ec
VZ
1714
1715 return NULL;
1716}
1717
1718void wxCSConv::CreateConvIfNeeded() const
1719{
1720 if ( m_deferred )
1721 {
1722 wxCSConv *self = (wxCSConv *)this; // const_cast
bda3d86a
VZ
1723
1724#if wxUSE_INTL
1725 // if we don't have neither the name nor the encoding, use the default
1726 // encoding for this system
1727 if ( !m_name && m_encoding == wxFONTENCODING_SYSTEM )
1728 {
4d312c22 1729 self->m_name = wxStrdup(wxLocale::GetSystemEncodingName());
bda3d86a
VZ
1730 }
1731#endif // wxUSE_INTL
1732
e95354ec
VZ
1733 self->m_convReal = DoCreate();
1734 self->m_deferred = false;
6001e347 1735 }
6001e347
RR
1736}
1737
1738size_t wxCSConv::MB2WC(wchar_t *buf, const char *psz, size_t n) const
1739{
e95354ec 1740 CreateConvIfNeeded();
dccce9ea 1741
e95354ec
VZ
1742 if (m_convReal)
1743 return m_convReal->MB2WC(buf, psz, n);
f1339c56
RR
1744
1745 // latin-1 (direct)
4def3b35 1746 size_t len = strlen(psz);
dccce9ea 1747
f1339c56
RR
1748 if (buf)
1749 {
4def3b35 1750 for (size_t c = 0; c <= len; c++)
f1339c56
RR
1751 buf[c] = (unsigned char)(psz[c]);
1752 }
dccce9ea 1753
f1339c56 1754 return len;
6001e347
RR
1755}
1756
1757size_t wxCSConv::WC2MB(char *buf, const wchar_t *psz, size_t n) const
1758{
e95354ec 1759 CreateConvIfNeeded();
dccce9ea 1760
e95354ec
VZ
1761 if (m_convReal)
1762 return m_convReal->WC2MB(buf, psz, n);
1cd52418 1763
f1339c56 1764 // latin-1 (direct)
f8d791e0 1765 const size_t len = wxWcslen(psz);
f1339c56
RR
1766 if (buf)
1767 {
4def3b35 1768 for (size_t c = 0; c <= len; c++)
24642831
VS
1769 {
1770 if (psz[c] > 0xFF)
1771 return (size_t)-1;
1772 buf[c] = psz[c];
1773 }
1774 }
1775 else
1776 {
1777 for (size_t c = 0; c <= len; c++)
1778 {
1779 if (psz[c] > 0xFF)
1780 return (size_t)-1;
1781 }
f1339c56 1782 }
dccce9ea 1783
f1339c56 1784 return len;
6001e347
RR
1785}
1786
bde4baac
VZ
1787// ----------------------------------------------------------------------------
1788// globals
1789// ----------------------------------------------------------------------------
1790
1791#ifdef __WINDOWS__
1792 static wxMBConv_win32 wxConvLibcObj;
f81f5901
SC
1793#elif defined(__WXMAC__) && !defined(__MACH__)
1794 static wxMBConv_mac wxConvLibcObj ;
bde4baac 1795#else
dcc8fac0 1796 static wxMBConvLibc wxConvLibcObj;
bde4baac
VZ
1797#endif
1798
1799static wxCSConv wxConvLocalObj(wxFONTENCODING_SYSTEM);
1800static wxCSConv wxConvISO8859_1Obj(wxFONTENCODING_ISO8859_1);
1801static wxMBConvUTF7 wxConvUTF7Obj;
1802static wxMBConvUTF8 wxConvUTF8Obj;
1803
1804
1805WXDLLIMPEXP_DATA_BASE(wxMBConv&) wxConvLibc = wxConvLibcObj;
1806WXDLLIMPEXP_DATA_BASE(wxCSConv&) wxConvLocal = wxConvLocalObj;
1807WXDLLIMPEXP_DATA_BASE(wxCSConv&) wxConvISO8859_1 = wxConvISO8859_1Obj;
1808WXDLLIMPEXP_DATA_BASE(wxMBConvUTF7&) wxConvUTF7 = wxConvUTF7Obj;
1809WXDLLIMPEXP_DATA_BASE(wxMBConvUTF8&) wxConvUTF8 = wxConvUTF8Obj;
1810WXDLLIMPEXP_DATA_BASE(wxMBConv *) wxConvCurrent = &wxConvLibcObj;
1811
1812#else // !wxUSE_WCHAR_T
1813
1814// stand-ins in absence of wchar_t
1815WXDLLIMPEXP_DATA_BASE(wxMBConv) wxConvLibc,
1816 wxConvISO8859_1,
1817 wxConvLocal,
1818 wxConvUTF8;
1819
1820#endif // wxUSE_WCHAR_T/!wxUSE_WCHAR_T
6001e347
RR
1821
1822