projects / wxWidgets.git / blame_incremental
? search:
summary | shortlog | log | commit | commitdiff | tree
blob | blame (non-incremental) | history | HEAD
Moved code around preparing the sorted model.
[wxWidgets.git] / src / common / strconv.cpp
0 / 2856 (  0%)
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// ============================================================================
16// declarations
17// ============================================================================
18
19// ----------------------------------------------------------------------------
20// headers
21// ----------------------------------------------------------------------------
22
23// For compilers that support precompilation, includes "wx.h".
24#include "wx/wxprec.h"
25
26#ifdef __BORLANDC__
27 #pragma hdrstop
28#endif
29
30#ifndef WX_PRECOMP
31 #include "wx/intl.h"
32 #include "wx/log.h"
33#endif // WX_PRECOMP
34
35#include "wx/strconv.h"
36
37#if wxUSE_WCHAR_T
38
39#ifdef __WINDOWS__
40 #include "wx/msw/private.h"
41 #include "wx/msw/missing.h"
42#endif
43
44#ifndef __WXWINCE__
45#include <errno.h>
46#endif
47
48#include <ctype.h>
49#include <string.h>
50#include <stdlib.h>
51
52#if defined(__WIN32__) && !defined(__WXMICROWIN__)
53 #define wxHAVE_WIN32_MB2WC
54#endif // __WIN32__ but !__WXMICROWIN__
55
56#ifdef __SALFORDC__
57 #include <clib.h>
58#endif
59
60#ifdef HAVE_ICONV
61 #include <iconv.h>
62 #include "wx/thread.h"
63#endif
64
65#include "wx/encconv.h"
66#include "wx/fontmap.h"
67#include "wx/utils.h"
68
69#ifdef __WXMAC__
70#ifndef __DARWIN__
71#include <ATSUnicode.h>
72#include <TextCommon.h>
73#include <TextEncodingConverter.h>
74#endif
75
76#include "wx/mac/private.h" // includes mac headers
77#endif
78
79#define TRACE_STRCONV _T("strconv")
80
81#if SIZEOF_WCHAR_T == 2
82 #define WC_UTF16
83#endif
84
85// ============================================================================
86// implementation
87// ============================================================================
88
89// ----------------------------------------------------------------------------
90// UTF-16 en/decoding to/from UCS-4
91// ----------------------------------------------------------------------------
92
93
94static size_t encode_utf16(wxUint32 input, wxUint16 *output)
95{
96 if (input<=0xffff)
97 {
98 if (output)
99 *output = (wxUint16) input;
100 return 1;
101 }
102 else if (input>=0x110000)
103 {
104 return (size_t)-1;
105 }
106 else
107 {
108 if (output)
109 {
110 *output++ = (wxUint16) ((input >> 10)+0xd7c0);
111 *output = (wxUint16) ((input&0x3ff)+0xdc00);
112 }
113 return 2;
114 }
115}
116
117static size_t decode_utf16(const wxUint16* input, wxUint32& output)
118{
119 if ((*input<0xd800) || (*input>0xdfff))
120 {
121 output = *input;
122 return 1;
123 }
124 else if ((input[1]<0xdc00) || (input[1]>0xdfff))
125 {
126 output = *input;
127 return (size_t)-1;
128 }
129 else
130 {
131 output = ((input[0] - 0xd7c0) << 10) + (input[1] - 0xdc00);
132 return 2;
133 }
134}
135
136
137// ----------------------------------------------------------------------------
138// wxMBConv
139// ----------------------------------------------------------------------------
140
141wxMBConv::~wxMBConv()
142{
143 // nothing to do here (necessary for Darwin linking probably)
144}
145
146const wxWCharBuffer wxMBConv::cMB2WC(const char *psz) const
147{
148 if ( psz )
149 {
150 // calculate the length of the buffer needed first
151 size_t nLen = MB2WC(NULL, psz, 0);
152 if ( nLen != (size_t)-1 )
153 {
154 // now do the actual conversion
155 wxWCharBuffer buf(nLen);
156 nLen = MB2WC(buf.data(), psz, nLen + 1); // with the trailing NULL
157 if ( nLen != (size_t)-1 )
158 {
159 return buf;
160 }
161 }
162 }
163
164 wxWCharBuffer buf((wchar_t *)NULL);
165
166 return buf;
167}
168
169const wxCharBuffer wxMBConv::cWC2MB(const wchar_t *pwz) const
170{
171 if ( pwz )
172 {
173 size_t nLen = WC2MB(NULL, pwz, 0);
174 if ( nLen != (size_t)-1 )
175 {
176 wxCharBuffer buf(nLen+3); // space for a wxUint32 trailing zero
177 nLen = WC2MB(buf.data(), pwz, nLen + 4);
178 if ( nLen != (size_t)-1 )
179 {
180 return buf;
181 }
182 }
183 }
184
185 wxCharBuffer buf((char *)NULL);
186
187 return buf;
188}
189
190const wxWCharBuffer wxMBConv::cMB2WC(const char *szString, size_t nStringLen, size_t* pOutSize) const
191{
192 wxASSERT(pOutSize != NULL);
193
194 const char* szEnd = szString + nStringLen + 1;
195 const char* szPos = szString;
196 const char* szStart = szPos;
197
198 size_t nActualLength = 0;
199 size_t nCurrentSize = nStringLen; //try normal size first (should never resize?)
200
201 wxWCharBuffer theBuffer(nCurrentSize);
202
203 //Convert the string until the length() is reached, continuing the
204 //loop every time a null character is reached
205 while(szPos != szEnd)
206 {
207 wxASSERT(szPos < szEnd); //something is _really_ screwed up if this rings true
208
209 //Get the length of the current (sub)string
210 size_t nLen = MB2WC(NULL, szPos, 0);
211
212 //Invalid conversion?
213 if( nLen == (size_t)-1 )
214 {
215 *pOutSize = 0;
216 theBuffer.data()[0u] = wxT('\0');
217 return theBuffer;
218 }
219
220
221 //Increase the actual length (+1 for current null character)
222 nActualLength += nLen + 1;
223
224 //if buffer too big, realloc the buffer
225 if (nActualLength > (nCurrentSize+1))
226 {
227 wxWCharBuffer theNewBuffer(nCurrentSize << 1);
228 memcpy(theNewBuffer.data(), theBuffer.data(), nCurrentSize * sizeof(wchar_t));
229 theBuffer = theNewBuffer;
230 nCurrentSize <<= 1;
231 }
232
233 //Convert the current (sub)string
234 if ( MB2WC(&theBuffer.data()[szPos - szStart], szPos, nLen + 1) == (size_t)-1 )
235 {
236 *pOutSize = 0;
237 theBuffer.data()[0u] = wxT('\0');
238 return theBuffer;
239 }
240
241 //Increment to next (sub)string
242 //Note that we have to use strlen instead of nLen here
243 //because XX2XX gives us the size of the output buffer,
244 //which is not necessarily the length of the string
245 szPos += strlen(szPos) + 1;
246 }
247
248 //success - return actual length and the buffer
249 *pOutSize = nActualLength;
250 return theBuffer;
251}
252
253const wxCharBuffer wxMBConv::cWC2MB(const wchar_t *szString, size_t nStringLen, size_t* pOutSize) const
254{
255 wxASSERT(pOutSize != NULL);
256
257 const wchar_t* szEnd = szString + nStringLen + 1;
258 const wchar_t* szPos = szString;
259 const wchar_t* szStart = szPos;
260
261 size_t nActualLength = 0;
262 size_t nCurrentSize = nStringLen << 2; //try * 4 first
263
264 wxCharBuffer theBuffer(nCurrentSize);
265
266 //Convert the string until the length() is reached, continuing the
267 //loop every time a null character is reached
268 while(szPos != szEnd)
269 {
270 wxASSERT(szPos < szEnd); //something is _really_ screwed up if this rings true
271
272 //Get the length of the current (sub)string
273 size_t nLen = WC2MB(NULL, szPos, 0);
274
275 //Invalid conversion?
276 if( nLen == (size_t)-1 )
277 {
278 *pOutSize = 0;
279 theBuffer.data()[0u] = wxT('\0');
280 return theBuffer;
281 }
282
283 //Increase the actual length (+1 for current null character)
284 nActualLength += nLen + 1;
285
286 //if buffer too big, realloc the buffer
287 if (nActualLength > (nCurrentSize+1))
288 {
289 wxCharBuffer theNewBuffer(nCurrentSize << 1);
290 memcpy(theNewBuffer.data(), theBuffer.data(), nCurrentSize);
291 theBuffer = theNewBuffer;
292 nCurrentSize <<= 1;
293 }
294
295 //Convert the current (sub)string
296 if(WC2MB(&theBuffer.data()[szPos - szStart], szPos, nLen + 1) == (size_t)-1 )
297 {
298 *pOutSize = 0;
299 theBuffer.data()[0u] = wxT('\0');
300 return theBuffer;
301 }
302
303 //Increment to next (sub)string
304 //Note that we have to use wxWcslen instead of nLen here
305 //because XX2XX gives us the size of the output buffer,
306 //which is not necessarily the length of the string
307 szPos += wxWcslen(szPos) + 1;
308 }
309
310 //success - return actual length and the buffer
311 *pOutSize = nActualLength;
312 return theBuffer;
313}
314
315// ----------------------------------------------------------------------------
316// wxMBConvLibc
317// ----------------------------------------------------------------------------
318
319size_t wxMBConvLibc::MB2WC(wchar_t *buf, const char *psz, size_t n) const
320{
321 return wxMB2WC(buf, psz, n);
322}
323
324size_t wxMBConvLibc::WC2MB(char *buf, const wchar_t *psz, size_t n) const
325{
326 return wxWC2MB(buf, psz, n);
327}
328
329#ifdef __UNIX__
330
331// ----------------------------------------------------------------------------
332// wxConvBrokenFileNames
333// ----------------------------------------------------------------------------
334
335wxConvBrokenFileNames::wxConvBrokenFileNames(const wxChar *charset)
336{
337 if ( !charset || wxStricmp(charset, _T("UTF-8")) == 0
338 || wxStricmp(charset, _T("UTF8")) == 0 )
339 m_conv = new wxMBConvUTF8(wxMBConvUTF8::MAP_INVALID_UTF8_TO_OCTAL);
340 else
341 m_conv = new wxCSConv(charset);
342}
343
344size_t
345wxConvBrokenFileNames::MB2WC(wchar_t *outputBuf,
346 const char *psz,
347 size_t outputSize) const
348{
349 return m_conv->MB2WC( outputBuf, psz, outputSize );
350}
351
352size_t
353wxConvBrokenFileNames::WC2MB(char *outputBuf,
354 const wchar_t *psz,
355 size_t outputSize) const
356{
357 return m_conv->WC2MB( outputBuf, psz, outputSize );
358}
359
360#endif
361
362// ----------------------------------------------------------------------------
363// UTF-7
364// ----------------------------------------------------------------------------
365
366// Implementation (C) 2004 Fredrik Roubert
367
368//
369// BASE64 decoding table
370//
371static const unsigned char utf7unb64[] =
372{
373 0xff, 0xff, 0xff, 0xff, 0xff, 0xff, 0xff, 0xff,
374 0xff, 0xff, 0xff, 0xff, 0xff, 0xff, 0xff, 0xff,
375 0xff, 0xff, 0xff, 0xff, 0xff, 0xff, 0xff, 0xff,
376 0xff, 0xff, 0xff, 0xff, 0xff, 0xff, 0xff, 0xff,
377 0xff, 0xff, 0xff, 0xff, 0xff, 0xff, 0xff, 0xff,
378 0xff, 0xff, 0xff, 0x3e, 0xff, 0xff, 0xff, 0x3f,
379 0x34, 0x35, 0x36, 0x37, 0x38, 0x39, 0x3a, 0x3b,
380 0x3c, 0x3d, 0xff, 0xff, 0xff, 0xff, 0xff, 0xff,
381 0xff, 0x00, 0x01, 0x02, 0x03, 0x04, 0x05, 0x06,
382 0x07, 0x08, 0x09, 0x0a, 0x0b, 0x0c, 0x0d, 0x0e,
383 0x0f, 0x10, 0x11, 0x12, 0x13, 0x14, 0x15, 0x16,
384 0x17, 0x18, 0x19, 0xff, 0xff, 0xff, 0xff, 0xff,
385 0xff, 0x1a, 0x1b, 0x1c, 0x1d, 0x1e, 0x1f, 0x20,
386 0x21, 0x22, 0x23, 0x24, 0x25, 0x26, 0x27, 0x28,
387 0x29, 0x2a, 0x2b, 0x2c, 0x2d, 0x2e, 0x2f, 0x30,
388 0x31, 0x32, 0x33, 0xff, 0xff, 0xff, 0xff, 0xff,
389 0xff, 0xff, 0xff, 0xff, 0xff, 0xff, 0xff, 0xff,
390 0xff, 0xff, 0xff, 0xff, 0xff, 0xff, 0xff, 0xff,
391 0xff, 0xff, 0xff, 0xff, 0xff, 0xff, 0xff, 0xff,
392 0xff, 0xff, 0xff, 0xff, 0xff, 0xff, 0xff, 0xff,
393 0xff, 0xff, 0xff, 0xff, 0xff, 0xff, 0xff, 0xff,
394 0xff, 0xff, 0xff, 0xff, 0xff, 0xff, 0xff, 0xff,
395 0xff, 0xff, 0xff, 0xff, 0xff, 0xff, 0xff, 0xff,
396 0xff, 0xff, 0xff, 0xff, 0xff, 0xff, 0xff, 0xff,
397 0xff, 0xff, 0xff, 0xff, 0xff, 0xff, 0xff, 0xff,
398 0xff, 0xff, 0xff, 0xff, 0xff, 0xff, 0xff, 0xff,
399 0xff, 0xff, 0xff, 0xff, 0xff, 0xff, 0xff, 0xff,
400 0xff, 0xff, 0xff, 0xff, 0xff, 0xff, 0xff, 0xff,
401 0xff, 0xff, 0xff, 0xff, 0xff, 0xff, 0xff, 0xff,
402 0xff, 0xff, 0xff, 0xff, 0xff, 0xff, 0xff, 0xff,
403 0xff, 0xff, 0xff, 0xff, 0xff, 0xff, 0xff, 0xff,
404 0xff, 0xff, 0xff, 0xff, 0xff, 0xff, 0xff, 0xff
405};
406
407size_t wxMBConvUTF7::MB2WC(wchar_t *buf, const char *psz, size_t n) const
408{
409 size_t len = 0;
410
411 while (*psz && ((!buf) || (len < n)))
412 {
413 unsigned char cc = *psz++;
414 if (cc != '+')
415 {
416 // plain ASCII char
417 if (buf)
418 *buf++ = cc;
419 len++;
420 }
421 else if (*psz == '-')
422 {
423 // encoded plus sign
424 if (buf)
425 *buf++ = cc;
426 len++;
427 psz++;
428 }
429 else
430 {
431 // BASE64 encoded string
432 bool lsb;
433 unsigned char c;
434 unsigned int d, l;
435 for (lsb = false, d = 0, l = 0;
436 (cc = utf7unb64[(unsigned char)*psz]) != 0xff; psz++)
437 {
438 d <<= 6;
439 d += cc;
440 for (l += 6; l >= 8; lsb = !lsb)
441 {
442 c = (unsigned char)((d >> (l -= 8)) % 256);
443 if (lsb)
444 {
445 if (buf)
446 *buf++ |= c;
447 len ++;
448 }
449 else
450 if (buf)
451 *buf = (wchar_t)(c << 8);
452 }
453 }
454 if (*psz == '-')
455 psz++;
456 }
457 }
458 if (buf && (len < n))
459 *buf = 0;
460 return len;
461}
462
463//
464// BASE64 encoding table
465//
466static const unsigned char utf7enb64[] =
467{
468 'A', 'B', 'C', 'D', 'E', 'F', 'G', 'H',
469 'I', 'J', 'K', 'L', 'M', 'N', 'O', 'P',
470 'Q', 'R', 'S', 'T', 'U', 'V', 'W', 'X',
471 'Y', 'Z', 'a', 'b', 'c', 'd', 'e', 'f',
472 'g', 'h', 'i', 'j', 'k', 'l', 'm', 'n',
473 'o', 'p', 'q', 'r', 's', 't', 'u', 'v',
474 'w', 'x', 'y', 'z', '0', '1', '2', '3',
475 '4', '5', '6', '7', '8', '9', '+', '/'
476};
477
478//
479// UTF-7 encoding table
480//
481// 0 - Set D (directly encoded characters)
482// 1 - Set O (optional direct characters)
483// 2 - whitespace characters (optional)
484// 3 - special characters
485//
486static const unsigned char utf7encode[128] =
487{
488 3, 3, 3, 3, 3, 3, 3, 3, 3, 2, 2, 3, 3, 2, 3, 3,
489 3, 3, 3, 3, 3, 3, 3, 3, 3, 3, 3, 3, 3, 3, 3, 3,
490 2, 1, 1, 1, 1, 1, 1, 0, 0, 0, 1, 3, 0, 0, 0, 3,
491 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 1, 1, 1, 1, 0,
492 1, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0,
493 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 1, 3, 1, 1, 1,
494 1, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0,
495 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 1, 1, 1, 3, 3
496};
497
498size_t wxMBConvUTF7::WC2MB(char *buf, const wchar_t *psz, size_t n) const
499{
500
501
502 size_t len = 0;
503
504 while (*psz && ((!buf) || (len < n)))
505 {
506 wchar_t cc = *psz++;
507 if (cc < 0x80 && utf7encode[cc] < 1)
508 {
509 // plain ASCII char
510 if (buf)
511 *buf++ = (char)cc;
512 len++;
513 }
514#ifndef WC_UTF16
515 else if (((wxUint32)cc) > 0xffff)
516 {
517 // no surrogate pair generation (yet?)
518 return (size_t)-1;
519 }
520#endif
521 else
522 {
523 if (buf)
524 *buf++ = '+';
525 len++;
526 if (cc != '+')
527 {
528 // BASE64 encode string
529 unsigned int lsb, d, l;
530 for (d = 0, l = 0; /*nothing*/; psz++)
531 {
532 for (lsb = 0; lsb < 2; lsb ++)
533 {
534 d <<= 8;
535 d += lsb ? cc & 0xff : (cc & 0xff00) >> 8;
536
537 for (l += 8; l >= 6; )
538 {
539 l -= 6;
540 if (buf)
541 *buf++ = utf7enb64[(d >> l) % 64];
542 len++;
543 }
544 }
545 cc = *psz;
546 if (!(cc) || (cc < 0x80 && utf7encode[cc] < 1))
547 break;
548 }
549 if (l != 0)
550 {
551 if (buf)
552 *buf++ = utf7enb64[((d % 16) << (6 - l)) % 64];
553 len++;
554 }
555 }
556 if (buf)
557 *buf++ = '-';
558 len++;
559 }
560 }
561 if (buf && (len < n))
562 *buf = 0;
563 return len;
564}
565
566// ----------------------------------------------------------------------------
567// UTF-8
568// ----------------------------------------------------------------------------
569
570static wxUint32 utf8_max[]=
571 { 0x7f, 0x7ff, 0xffff, 0x1fffff, 0x3ffffff, 0x7fffffff, 0xffffffff };
572
573// boundaries of the private use area we use to (temporarily) remap invalid
574// characters invalid in a UTF-8 encoded string
575const wxUint32 wxUnicodePUA = 0x100000;
576const wxUint32 wxUnicodePUAEnd = wxUnicodePUA + 256;
577
578size_t wxMBConvUTF8::MB2WC(wchar_t *buf, const char *psz, size_t n) const
579{
580 size_t len = 0;
581
582 while (*psz && ((!buf) || (len < n)))
583 {
584 const char *opsz = psz;
585 bool invalid = false;
586 unsigned char cc = *psz++, fc = cc;
587 unsigned cnt;
588 for (cnt = 0; fc & 0x80; cnt++)
589 fc <<= 1;
590 if (!cnt)
591 {
592 // plain ASCII char
593 if (buf)
594 *buf++ = cc;
595 len++;
596
597 // escape the escape character for octal escapes
598 if ((m_options & MAP_INVALID_UTF8_TO_OCTAL)
599 && cc == '\\' && (!buf || len < n))
600 {
601 if (buf)
602 *buf++ = cc;
603 len++;
604 }
605 }
606 else
607 {
608 cnt--;
609 if (!cnt)
610 {
611 // invalid UTF-8 sequence
612 invalid = true;
613 }
614 else
615 {
616 unsigned ocnt = cnt - 1;
617 wxUint32 res = cc & (0x3f >> cnt);
618 while (cnt--)
619 {
620 cc = *psz;
621 if ((cc & 0xC0) != 0x80)
622 {
623 // invalid UTF-8 sequence
624 invalid = true;
625 break;
626 }
627 psz++;
628 res = (res << 6) | (cc & 0x3f);
629 }
630 if (invalid || res <= utf8_max[ocnt])
631 {
632 // illegal UTF-8 encoding
633 invalid = true;
634 }
635 else if ((m_options & MAP_INVALID_UTF8_TO_PUA) &&
636 res >= wxUnicodePUA && res < wxUnicodePUAEnd)
637 {
638 // if one of our PUA characters turns up externally
639 // it must also be treated as an illegal sequence
640 // (a bit like you have to escape an escape character)
641 invalid = true;
642 }
643 else
644 {
645#ifdef WC_UTF16
646 // cast is ok because wchar_t == wxUuint16 if WC_UTF16
647 size_t pa = encode_utf16(res, (wxUint16 *)buf);
648 if (pa == (size_t)-1)
649 {
650 invalid = true;
651 }
652 else
653 {
654 if (buf)
655 buf += pa;
656 len += pa;
657 }
658#else // !WC_UTF16
659 if (buf)
660 *buf++ = (wchar_t)res;
661 len++;
662#endif // WC_UTF16/!WC_UTF16
663 }
664 }
665 if (invalid)
666 {
667 if (m_options & MAP_INVALID_UTF8_TO_PUA)
668 {
669 while (opsz < psz && (!buf || len < n))
670 {
671#ifdef WC_UTF16
672 // cast is ok because wchar_t == wxUuint16 if WC_UTF16
673 size_t pa = encode_utf16((unsigned char)*opsz + wxUnicodePUA, (wxUint16 *)buf);
674 wxASSERT(pa != (size_t)-1);
675 if (buf)
676 buf += pa;
677 opsz++;
678 len += pa;
679#else
680 if (buf)
681 *buf++ = (wchar_t)(wxUnicodePUA + (unsigned char)*opsz);
682 opsz++;
683 len++;
684#endif
685 }
686 }
687 else if (m_options & MAP_INVALID_UTF8_TO_OCTAL)
688 {
689 while (opsz < psz && (!buf || len < n))
690 {
691 if ( buf && len + 3 < n )
692 {
693 unsigned char on = *opsz;
694 *buf++ = L'\\';
695 *buf++ = (wchar_t)( L'0' + on / 0100 );
696 *buf++ = (wchar_t)( L'0' + (on % 0100) / 010 );
697 *buf++ = (wchar_t)( L'0' + on % 010 );
698 }
699 opsz++;
700 len += 4;
701 }
702 }
703 else // MAP_INVALID_UTF8_NOT
704 {
705 return (size_t)-1;
706 }
707 }
708 }
709 }
710 if (buf && (len < n))
711 *buf = 0;
712 return len;
713}
714
715static inline bool isoctal(wchar_t wch)
716{
717 return L'0' <= wch && wch <= L'7';
718}
719
720size_t wxMBConvUTF8::WC2MB(char *buf, const wchar_t *psz, size_t n) const
721{
722 size_t len = 0;
723
724 while (*psz && ((!buf) || (len < n)))
725 {
726 wxUint32 cc;
727#ifdef WC_UTF16
728 // cast is ok for WC_UTF16
729 size_t pa = decode_utf16((const wxUint16 *)psz, cc);
730 psz += (pa == (size_t)-1) ? 1 : pa;
731#else
732 cc=(*psz++) & 0x7fffffff;
733#endif
734
735 if ( (m_options & MAP_INVALID_UTF8_TO_PUA)
736 && cc >= wxUnicodePUA && cc < wxUnicodePUAEnd )
737 {
738 if (buf)
739 *buf++ = (char)(cc - wxUnicodePUA);
740 len++;
741 }
742 else if ( (m_options & MAP_INVALID_UTF8_TO_OCTAL)
743 && cc == L'\\' && psz[0] == L'\\' )
744 {
745 if (buf)
746 *buf++ = (char)cc;
747 psz++;
748 len++;
749 }
750 else if ( (m_options & MAP_INVALID_UTF8_TO_OCTAL) &&
751 cc == L'\\' &&
752 isoctal(psz[0]) && isoctal(psz[1]) && isoctal(psz[2]) )
753 {
754 if (buf)
755 {
756 *buf++ = (char) ((psz[0] - L'0')*0100 +
757 (psz[1] - L'0')*010 +
758 (psz[2] - L'0'));
759 }
760
761 psz += 3;
762 len++;
763 }
764 else
765 {
766 unsigned cnt;
767 for (cnt = 0; cc > utf8_max[cnt]; cnt++) {}
768 if (!cnt)
769 {
770 // plain ASCII char
771 if (buf)
772 *buf++ = (char) cc;
773 len++;
774 }
775
776 else
777 {
778 len += cnt + 1;
779 if (buf)
780 {
781 *buf++ = (char) ((-128 >> cnt) | ((cc >> (cnt * 6)) & (0x3f >> cnt)));
782 while (cnt--)
783 *buf++ = (char) (0x80 | ((cc >> (cnt * 6)) & 0x3f));
784 }
785 }
786 }
787 }
788
789 if (buf && (len<n))
790 *buf = 0;
791
792 return len;
793}
794
795// ----------------------------------------------------------------------------
796// UTF-16
797// ----------------------------------------------------------------------------
798
799#ifdef WORDS_BIGENDIAN
800 #define wxMBConvUTF16straight wxMBConvUTF16BE
801 #define wxMBConvUTF16swap wxMBConvUTF16LE
802#else
803 #define wxMBConvUTF16swap wxMBConvUTF16BE
804 #define wxMBConvUTF16straight wxMBConvUTF16LE
805#endif
806
807
808#ifdef WC_UTF16
809
810// copy 16bit MB to 16bit String
811size_t wxMBConvUTF16straight::MB2WC(wchar_t *buf, const char *psz, size_t n) const
812{
813 size_t len=0;
814
815 while (*(wxUint16*)psz && (!buf || len < n))
816 {
817 if (buf)
818 *buf++ = *(wxUint16*)psz;
819 len++;
820
821 psz += sizeof(wxUint16);
822 }
823 if (buf && len<n) *buf=0;
824
825 return len;
826}
827
828
829// copy 16bit String to 16bit MB
830size_t wxMBConvUTF16straight::WC2MB(char *buf, const wchar_t *psz, size_t n) const
831{
832 size_t len=0;
833
834 while (*psz && (!buf || len < n))
835 {
836 if (buf)
837 {
838 *(wxUint16*)buf = *psz;
839 buf += sizeof(wxUint16);
840 }
841 len += sizeof(wxUint16);
842 psz++;
843 }
844 if (buf && len<=n-sizeof(wxUint16)) *(wxUint16*)buf=0;
845
846 return len;
847}
848
849
850// swap 16bit MB to 16bit String
851size_t wxMBConvUTF16swap::MB2WC(wchar_t *buf, const char *psz, size_t n) const
852{
853 size_t len=0;
854
855 while (*(wxUint16*)psz && (!buf || len < n))
856 {
857 if (buf)
858 {
859 ((char *)buf)[0] = psz[1];
860 ((char *)buf)[1] = psz[0];
861 buf++;
862 }
863 len++;
864 psz += sizeof(wxUint16);
865 }
866 if (buf && len<n) *buf=0;
867
868 return len;
869}
870
871
872// swap 16bit MB to 16bit String
873size_t wxMBConvUTF16swap::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 if (buf)
880 {
881 *buf++ = ((char*)psz)[1];
882 *buf++ = ((char*)psz)[0];
883 }
884 len += sizeof(wxUint16);
885 psz++;
886 }
887 if (buf && len<=n-sizeof(wxUint16)) *(wxUint16*)buf=0;
888
889 return len;
890}
891
892
893#else // WC_UTF16
894
895
896// copy 16bit MB to 32bit String
897size_t wxMBConvUTF16straight::MB2WC(wchar_t *buf, const char *psz, size_t n) const
898{
899 size_t len=0;
900
901 while (*(wxUint16*)psz && (!buf || len < n))
902 {
903 wxUint32 cc;
904 size_t pa=decode_utf16((wxUint16*)psz, cc);
905 if (pa == (size_t)-1)
906 return pa;
907
908 if (buf)
909 *buf++ = (wchar_t)cc;
910 len++;
911 psz += pa * sizeof(wxUint16);
912 }
913 if (buf && len<n) *buf=0;
914
915 return len;
916}
917
918
919// copy 32bit String to 16bit MB
920size_t wxMBConvUTF16straight::WC2MB(char *buf, const wchar_t *psz, size_t n) const
921{
922 size_t len=0;
923
924 while (*psz && (!buf || len < n))
925 {
926 wxUint16 cc[2];
927 size_t pa=encode_utf16(*psz, cc);
928
929 if (pa == (size_t)-1)
930 return pa;
931
932 if (buf)
933 {
934 *(wxUint16*)buf = cc[0];
935 buf += sizeof(wxUint16);
936 if (pa > 1)
937 {
938 *(wxUint16*)buf = cc[1];
939 buf += sizeof(wxUint16);
940 }
941 }
942
943 len += pa*sizeof(wxUint16);
944 psz++;
945 }
946 if (buf && len<=n-sizeof(wxUint16)) *(wxUint16*)buf=0;
947
948 return len;
949}
950
951
952// swap 16bit MB to 32bit String
953size_t wxMBConvUTF16swap::MB2WC(wchar_t *buf, const char *psz, size_t n) const
954{
955 size_t len=0;
956
957 while (*(wxUint16*)psz && (!buf || len < n))
958 {
959 wxUint32 cc;
960 char tmp[4];
961 tmp[0]=psz[1]; tmp[1]=psz[0];
962 tmp[2]=psz[3]; tmp[3]=psz[2];
963
964 size_t pa=decode_utf16((wxUint16*)tmp, cc);
965 if (pa == (size_t)-1)
966 return pa;
967
968 if (buf)
969 *buf++ = (wchar_t)cc;
970
971 len++;
972 psz += pa * sizeof(wxUint16);
973 }
974 if (buf && len<n) *buf=0;
975
976 return len;
977}
978
979
980// swap 32bit String to 16bit MB
981size_t wxMBConvUTF16swap::WC2MB(char *buf, const wchar_t *psz, size_t n) const
982{
983 size_t len=0;
984
985 while (*psz && (!buf || len < n))
986 {
987 wxUint16 cc[2];
988 size_t pa=encode_utf16(*psz, cc);
989
990 if (pa == (size_t)-1)
991 return pa;
992
993 if (buf)
994 {
995 *buf++ = ((char*)cc)[1];
996 *buf++ = ((char*)cc)[0];
997 if (pa > 1)
998 {
999 *buf++ = ((char*)cc)[3];
1000 *buf++ = ((char*)cc)[2];
1001 }
1002 }
1003
1004 len += pa*sizeof(wxUint16);
1005 psz++;
1006 }
1007 if (buf && len<=n-sizeof(wxUint16)) *(wxUint16*)buf=0;
1008
1009 return len;
1010}
1011
1012#endif // WC_UTF16
1013
1014
1015// ----------------------------------------------------------------------------
1016// UTF-32
1017// ----------------------------------------------------------------------------
1018
1019#ifdef WORDS_BIGENDIAN
1020#define wxMBConvUTF32straight wxMBConvUTF32BE
1021#define wxMBConvUTF32swap wxMBConvUTF32LE
1022#else
1023#define wxMBConvUTF32swap wxMBConvUTF32BE
1024#define wxMBConvUTF32straight wxMBConvUTF32LE
1025#endif
1026
1027
1028WXDLLIMPEXP_DATA_BASE(wxMBConvUTF32LE) wxConvUTF32LE;
1029WXDLLIMPEXP_DATA_BASE(wxMBConvUTF32BE) wxConvUTF32BE;
1030
1031
1032#ifdef WC_UTF16
1033
1034// copy 32bit MB to 16bit String
1035size_t wxMBConvUTF32straight::MB2WC(wchar_t *buf, const char *psz, size_t n) const
1036{
1037 size_t len=0;
1038
1039 while (*(wxUint32*)psz && (!buf || len < n))
1040 {
1041 wxUint16 cc[2];
1042
1043 size_t pa=encode_utf16(*(wxUint32*)psz, cc);
1044 if (pa == (size_t)-1)
1045 return pa;
1046
1047 if (buf)
1048 {
1049 *buf++ = cc[0];
1050 if (pa > 1)
1051 *buf++ = cc[1];
1052 }
1053 len += pa;
1054 psz += sizeof(wxUint32);
1055 }
1056 if (buf && len<n) *buf=0;
1057
1058 return len;
1059}
1060
1061
1062// copy 16bit String to 32bit MB
1063size_t wxMBConvUTF32straight::WC2MB(char *buf, const wchar_t *psz, size_t n) const
1064{
1065 size_t len=0;
1066
1067 while (*psz && (!buf || len < n))
1068 {
1069 wxUint32 cc;
1070
1071 // cast is ok for WC_UTF16
1072 size_t pa = decode_utf16((const wxUint16 *)psz, cc);
1073 if (pa == (size_t)-1)
1074 return pa;
1075
1076 if (buf)
1077 {
1078 *(wxUint32*)buf = cc;
1079 buf += sizeof(wxUint32);
1080 }
1081 len += sizeof(wxUint32);
1082 psz += pa;
1083 }
1084
1085 if (buf && len<=n-sizeof(wxUint32))
1086 *(wxUint32*)buf=0;
1087
1088 return len;
1089}
1090
1091
1092
1093// swap 32bit MB to 16bit String
1094size_t wxMBConvUTF32swap::MB2WC(wchar_t *buf, const char *psz, size_t n) const
1095{
1096 size_t len=0;
1097
1098 while (*(wxUint32*)psz && (!buf || len < n))
1099 {
1100 char tmp[4];
1101 tmp[0] = psz[3]; tmp[1] = psz[2];
1102 tmp[2] = psz[1]; tmp[3] = psz[0];
1103
1104
1105 wxUint16 cc[2];
1106
1107 size_t pa=encode_utf16(*(wxUint32*)tmp, cc);
1108 if (pa == (size_t)-1)
1109 return pa;
1110
1111 if (buf)
1112 {
1113 *buf++ = cc[0];
1114 if (pa > 1)
1115 *buf++ = cc[1];
1116 }
1117 len += pa;
1118 psz += sizeof(wxUint32);
1119 }
1120
1121 if (buf && len<n)
1122 *buf=0;
1123
1124 return len;
1125}
1126
1127
1128// swap 16bit String to 32bit MB
1129size_t wxMBConvUTF32swap::WC2MB(char *buf, const wchar_t *psz, size_t n) const
1130{
1131 size_t len=0;
1132
1133 while (*psz && (!buf || len < n))
1134 {
1135 char cc[4];
1136
1137 // cast is ok for WC_UTF16
1138 size_t pa=decode_utf16((const wxUint16 *)psz, *(wxUint32*)cc);
1139 if (pa == (size_t)-1)
1140 return pa;
1141
1142 if (buf)
1143 {
1144 *buf++ = cc[3];
1145 *buf++ = cc[2];
1146 *buf++ = cc[1];
1147 *buf++ = cc[0];
1148 }
1149 len += sizeof(wxUint32);
1150 psz += pa;
1151 }
1152
1153 if (buf && len<=n-sizeof(wxUint32))
1154 *(wxUint32*)buf=0;
1155
1156 return len;
1157}
1158
1159#else // WC_UTF16
1160
1161
1162// copy 32bit MB to 32bit String
1163size_t wxMBConvUTF32straight::MB2WC(wchar_t *buf, const char *psz, size_t n) const
1164{
1165 size_t len=0;
1166
1167 while (*(wxUint32*)psz && (!buf || len < n))
1168 {
1169 if (buf)
1170 *buf++ = (wchar_t)(*(wxUint32*)psz);
1171 len++;
1172 psz += sizeof(wxUint32);
1173 }
1174
1175 if (buf && len<n)
1176 *buf=0;
1177
1178 return len;
1179}
1180
1181
1182// copy 32bit String to 32bit MB
1183size_t wxMBConvUTF32straight::WC2MB(char *buf, const wchar_t *psz, size_t n) const
1184{
1185 size_t len=0;
1186
1187 while (*psz && (!buf || len < n))
1188 {
1189 if (buf)
1190 {
1191 *(wxUint32*)buf = *psz;
1192 buf += sizeof(wxUint32);
1193 }
1194
1195 len += sizeof(wxUint32);
1196 psz++;
1197 }
1198
1199 if (buf && len<=n-sizeof(wxUint32))
1200 *(wxUint32*)buf=0;
1201
1202 return len;
1203}
1204
1205
1206// swap 32bit MB to 32bit String
1207size_t wxMBConvUTF32swap::MB2WC(wchar_t *buf, const char *psz, size_t n) const
1208{
1209 size_t len=0;
1210
1211 while (*(wxUint32*)psz && (!buf || len < n))
1212 {
1213 if (buf)
1214 {
1215 ((char *)buf)[0] = psz[3];
1216 ((char *)buf)[1] = psz[2];
1217 ((char *)buf)[2] = psz[1];
1218 ((char *)buf)[3] = psz[0];
1219 buf++;
1220 }
1221 len++;
1222 psz += sizeof(wxUint32);
1223 }
1224
1225 if (buf && len<n)
1226 *buf=0;
1227
1228 return len;
1229}
1230
1231
1232// swap 32bit String to 32bit MB
1233size_t wxMBConvUTF32swap::WC2MB(char *buf, const wchar_t *psz, size_t n) const
1234{
1235 size_t len=0;
1236
1237 while (*psz && (!buf || len < n))
1238 {
1239 if (buf)
1240 {
1241 *buf++ = ((char *)psz)[3];
1242 *buf++ = ((char *)psz)[2];
1243 *buf++ = ((char *)psz)[1];
1244 *buf++ = ((char *)psz)[0];
1245 }
1246 len += sizeof(wxUint32);
1247 psz++;
1248 }
1249
1250 if (buf && len<=n-sizeof(wxUint32))
1251 *(wxUint32*)buf=0;
1252
1253 return len;
1254}
1255
1256
1257#endif // WC_UTF16
1258
1259
1260// ============================================================================
1261// The classes doing conversion using the iconv_xxx() functions
1262// ============================================================================
1263
1264#ifdef HAVE_ICONV
1265
1266// VS: glibc 2.1.3 is broken in that iconv() conversion to/from UCS4 fails with
1267// E2BIG if output buffer is _exactly_ as big as needed. Such case is
1268// (unless there's yet another bug in glibc) the only case when iconv()
1269// returns with (size_t)-1 (which means error) and says there are 0 bytes
1270// left in the input buffer -- when _real_ error occurs,
1271// bytes-left-in-input buffer is non-zero. Hence, this alternative test for
1272// iconv() failure.
1273// [This bug does not appear in glibc 2.2.]
1274#if defined(__GLIBC__) && __GLIBC__ == 2 && __GLIBC_MINOR__ <= 1
1275#define ICONV_FAILED(cres, bufLeft) ((cres == (size_t)-1) && \
1276 (errno != E2BIG || bufLeft != 0))
1277#else
1278#define ICONV_FAILED(cres, bufLeft) (cres == (size_t)-1)
1279#endif
1280
1281#define ICONV_CHAR_CAST(x) ((ICONV_CONST char **)(x))
1282
1283#define ICONV_T_INVALID ((iconv_t)-1)
1284
1285#if SIZEOF_WCHAR_T == 4
1286 #define WC_BSWAP wxUINT32_SWAP_ALWAYS
1287 #define WC_ENC wxFONTENCODING_UTF32
1288#elif SIZEOF_WCHAR_T == 2
1289 #define WC_BSWAP wxUINT16_SWAP_ALWAYS
1290 #define WC_ENC wxFONTENCODING_UTF16
1291#else // sizeof(wchar_t) != 2 nor 4
1292 // does this ever happen?
1293 #error "Unknown sizeof(wchar_t): please report this to wx-dev@lists.wxwindows.org"
1294#endif
1295
1296// ----------------------------------------------------------------------------
1297// wxMBConv_iconv: encapsulates an iconv character set
1298// ----------------------------------------------------------------------------
1299
1300class wxMBConv_iconv : public wxMBConv
1301{
1302public:
1303 wxMBConv_iconv(const wxChar *name);
1304 virtual ~wxMBConv_iconv();
1305
1306 virtual size_t MB2WC(wchar_t *buf, const char *psz, size_t n) const;
1307 virtual size_t WC2MB(char *buf, const wchar_t *psz, size_t n) const;
1308
1309 bool IsOk() const
1310 { return (m2w != ICONV_T_INVALID) && (w2m != ICONV_T_INVALID); }
1311
1312protected:
1313 // the iconv handlers used to translate from multibyte to wide char and in
1314 // the other direction
1315 iconv_t m2w,
1316 w2m;
1317#if wxUSE_THREADS
1318 // guards access to m2w and w2m objects
1319 wxMutex m_iconvMutex;
1320#endif
1321
1322private:
1323 // the name (for iconv_open()) of a wide char charset -- if none is
1324 // available on this machine, it will remain NULL
1325 static wxString ms_wcCharsetName;
1326
1327 // true if the wide char encoding we use (i.e. ms_wcCharsetName) has
1328 // different endian-ness than the native one
1329 static bool ms_wcNeedsSwap;
1330};
1331
1332// make the constructor available for unit testing
1333WXDLLIMPEXP_BASE wxMBConv* new_wxMBConv_iconv( const wxChar* name )
1334{
1335 wxMBConv_iconv* result = new wxMBConv_iconv( name );
1336 if ( !result->IsOk() )
1337 {
1338 delete result;
1339 return 0;
1340 }
1341 return result;
1342}
1343
1344wxString wxMBConv_iconv::ms_wcCharsetName;
1345bool wxMBConv_iconv::ms_wcNeedsSwap = false;
1346
1347wxMBConv_iconv::wxMBConv_iconv(const wxChar *name)
1348{
1349 // iconv operates with chars, not wxChars, but luckily it uses only ASCII
1350 // names for the charsets
1351 const wxCharBuffer cname(wxString(name).ToAscii());
1352
1353 // check for charset that represents wchar_t:
1354 if ( ms_wcCharsetName.empty() )
1355 {
1356 wxLogTrace(TRACE_STRCONV, _T("Looking for wide char codeset:"));
1357
1358#if wxUSE_FONTMAP
1359 const wxChar **names = wxFontMapperBase::GetAllEncodingNames(WC_ENC);
1360#else // !wxUSE_FONTMAP
1361 static const wxChar *names[] =
1362 {
1363#if SIZEOF_WCHAR_T == 4
1364 _T("UCS-4"),
1365#elif SIZEOF_WCHAR_T = 2
1366 _T("UCS-2"),
1367#endif
1368 NULL
1369 };
1370#endif // wxUSE_FONTMAP/!wxUSE_FONTMAP
1371
1372 for ( ; *names && ms_wcCharsetName.empty(); ++names )
1373 {
1374 const wxString nameCS(*names);
1375
1376 // first try charset with explicit bytesex info (e.g. "UCS-4LE"):
1377 wxString nameXE(nameCS);
1378 #ifdef WORDS_BIGENDIAN
1379 nameXE += _T("BE");
1380 #else // little endian
1381 nameXE += _T("LE");
1382 #endif
1383
1384 wxLogTrace(TRACE_STRCONV, _T(" trying charset \"%s\""),
1385 nameXE.c_str());
1386
1387 m2w = iconv_open(nameXE.ToAscii(), cname);
1388 if ( m2w == ICONV_T_INVALID )
1389 {
1390 // try charset w/o bytesex info (e.g. "UCS4")
1391 wxLogTrace(TRACE_STRCONV, _T(" trying charset \"%s\""),
1392 nameCS.c_str());
1393 m2w = iconv_open(nameCS.ToAscii(), cname);
1394
1395 // and check for bytesex ourselves:
1396 if ( m2w != ICONV_T_INVALID )
1397 {
1398 char buf[2], *bufPtr;
1399 wchar_t wbuf[2], *wbufPtr;
1400 size_t insz, outsz;
1401 size_t res;
1402
1403 buf[0] = 'A';
1404 buf[1] = 0;
1405 wbuf[0] = 0;
1406 insz = 2;
1407 outsz = SIZEOF_WCHAR_T * 2;
1408 wbufPtr = wbuf;
1409 bufPtr = buf;
1410
1411 res = iconv(m2w, ICONV_CHAR_CAST(&bufPtr), &insz,
1412 (char**)&wbufPtr, &outsz);
1413
1414 if (ICONV_FAILED(res, insz))
1415 {
1416 wxLogLastError(wxT("iconv"));
1417 wxLogError(_("Conversion to charset '%s' doesn't work."),
1418 nameCS.c_str());
1419 }
1420 else // ok, can convert to this encoding, remember it
1421 {
1422 ms_wcCharsetName = nameCS;
1423 ms_wcNeedsSwap = wbuf[0] != (wchar_t)buf[0];
1424 }
1425 }
1426 }
1427 else // use charset not requiring byte swapping
1428 {
1429 ms_wcCharsetName = nameXE;
1430 }
1431 }
1432
1433 wxLogTrace(TRACE_STRCONV,
1434 wxT("iconv wchar_t charset is \"%s\"%s"),
1435 ms_wcCharsetName.empty() ? _T("<none>")
1436 : ms_wcCharsetName.c_str(),
1437 ms_wcNeedsSwap ? _T(" (needs swap)")
1438 : _T(""));
1439 }
1440 else // we already have ms_wcCharsetName
1441 {
1442 m2w = iconv_open(ms_wcCharsetName.ToAscii(), cname);
1443 }
1444
1445 if ( ms_wcCharsetName.empty() )
1446 {
1447 w2m = ICONV_T_INVALID;
1448 }
1449 else
1450 {
1451 w2m = iconv_open(cname, ms_wcCharsetName.ToAscii());
1452 if ( w2m == ICONV_T_INVALID )
1453 {
1454 wxLogTrace(TRACE_STRCONV,
1455 wxT("\"%s\" -> \"%s\" works but not the converse!?"),
1456 ms_wcCharsetName.c_str(), cname.data());
1457 }
1458 }
1459}
1460
1461wxMBConv_iconv::~wxMBConv_iconv()
1462{
1463 if ( m2w != ICONV_T_INVALID )
1464 iconv_close(m2w);
1465 if ( w2m != ICONV_T_INVALID )
1466 iconv_close(w2m);
1467}
1468
1469size_t wxMBConv_iconv::MB2WC(wchar_t *buf, const char *psz, size_t n) const
1470{
1471#if wxUSE_THREADS
1472 // NB: iconv() is MT-safe, but each thread must use it's own iconv_t handle.
1473 // Unfortunately there is a couple of global wxCSConv objects such as
1474 // wxConvLocal that are used all over wx code, so we have to make sure
1475 // the handle is used by at most one thread at the time. Otherwise
1476 // only a few wx classes would be safe to use from non-main threads
1477 // as MB<->WC conversion would fail "randomly".
1478 wxMutexLocker lock(wxConstCast(this, wxMBConv_iconv)->m_iconvMutex);
1479#endif
1480
1481 size_t inbuf = strlen(psz);
1482 size_t outbuf = n * SIZEOF_WCHAR_T;
1483 size_t res, cres;
1484 // VS: Use these instead of psz, buf because iconv() modifies its arguments:
1485 wchar_t *bufPtr = buf;
1486 const char *pszPtr = psz;
1487
1488 if (buf)
1489 {
1490 // have destination buffer, convert there
1491 cres = iconv(m2w,
1492 ICONV_CHAR_CAST(&pszPtr), &inbuf,
1493 (char**)&bufPtr, &outbuf);
1494 res = n - (outbuf / SIZEOF_WCHAR_T);
1495
1496 if (ms_wcNeedsSwap)
1497 {
1498 // convert to native endianness
1499 for ( unsigned i = 0; i < res; i++ )
1500 buf[n] = WC_BSWAP(buf[i]);
1501 }
1502
1503 // NB: iconv was given only strlen(psz) characters on input, and so
1504 // it couldn't convert the trailing zero. Let's do it ourselves
1505 // if there's some room left for it in the output buffer.
1506 if (res < n)
1507 buf[res] = 0;
1508 }
1509 else
1510 {
1511 // no destination buffer... convert using temp buffer
1512 // to calculate destination buffer requirement
1513 wchar_t tbuf[8];
1514 res = 0;
1515 do {
1516 bufPtr = tbuf;
1517 outbuf = 8*SIZEOF_WCHAR_T;
1518
1519 cres = iconv(m2w,
1520 ICONV_CHAR_CAST(&pszPtr), &inbuf,
1521 (char**)&bufPtr, &outbuf );
1522
1523 res += 8-(outbuf/SIZEOF_WCHAR_T);
1524 } while ((cres==(size_t)-1) && (errno==E2BIG));
1525 }
1526
1527 if (ICONV_FAILED(cres, inbuf))
1528 {
1529 //VS: it is ok if iconv fails, hence trace only
1530 wxLogTrace(TRACE_STRCONV, wxT("iconv failed: %s"), wxSysErrorMsg(wxSysErrorCode()));
1531 return (size_t)-1;
1532 }
1533
1534 return res;
1535}
1536
1537size_t wxMBConv_iconv::WC2MB(char *buf, const wchar_t *psz, size_t n) const
1538{
1539#if wxUSE_THREADS
1540 // NB: explained in MB2WC
1541 wxMutexLocker lock(wxConstCast(this, wxMBConv_iconv)->m_iconvMutex);
1542#endif
1543
1544 size_t inlen = wxWcslen(psz);
1545 size_t inbuf = inlen * SIZEOF_WCHAR_T;
1546 size_t outbuf = n;
1547 size_t res, cres;
1548
1549 wchar_t *tmpbuf = 0;
1550
1551 if (ms_wcNeedsSwap)
1552 {
1553 // need to copy to temp buffer to switch endianness
1554 // (doing WC_BSWAP twice on the original buffer won't help, as it
1555 // could be in read-only memory, or be accessed in some other thread)
1556 tmpbuf = (wchar_t *)malloc(inbuf + SIZEOF_WCHAR_T);
1557 for ( size_t i = 0; i < inlen; i++ )
1558 tmpbuf[n] = WC_BSWAP(psz[i]);
1559 tmpbuf[inlen] = L'\0';
1560 psz = tmpbuf;
1561 }
1562
1563 if (buf)
1564 {
1565 // have destination buffer, convert there
1566 cres = iconv( w2m, ICONV_CHAR_CAST(&psz), &inbuf, &buf, &outbuf );
1567
1568 res = n-outbuf;
1569
1570 // NB: iconv was given only wcslen(psz) characters on input, and so
1571 // it couldn't convert the trailing zero. Let's do it ourselves
1572 // if there's some room left for it in the output buffer.
1573 if (res < n)
1574 buf[0] = 0;
1575 }
1576 else
1577 {
1578 // no destination buffer... convert using temp buffer
1579 // to calculate destination buffer requirement
1580 char tbuf[16];
1581 res = 0;
1582 do {
1583 buf = tbuf; outbuf = 16;
1584
1585 cres = iconv( w2m, ICONV_CHAR_CAST(&psz), &inbuf, &buf, &outbuf );
1586
1587 res += 16 - outbuf;
1588 } while ((cres==(size_t)-1) && (errno==E2BIG));
1589 }
1590
1591 if (ms_wcNeedsSwap)
1592 {
1593 free(tmpbuf);
1594 }
1595
1596 if (ICONV_FAILED(cres, inbuf))
1597 {
1598 wxLogTrace(TRACE_STRCONV, wxT("iconv failed: %s"), wxSysErrorMsg(wxSysErrorCode()));
1599 return (size_t)-1;
1600 }
1601
1602 return res;
1603}
1604
1605#endif // HAVE_ICONV
1606
1607
1608// ============================================================================
1609// Win32 conversion classes
1610// ============================================================================
1611
1612#ifdef wxHAVE_WIN32_MB2WC
1613
1614// from utils.cpp
1615#if wxUSE_FONTMAP
1616extern WXDLLIMPEXP_BASE long wxCharsetToCodepage(const wxChar *charset);
1617extern WXDLLIMPEXP_BASE long wxEncodingToCodepage(wxFontEncoding encoding);
1618#endif
1619
1620class wxMBConv_win32 : public wxMBConv
1621{
1622public:
1623 wxMBConv_win32()
1624 {
1625 m_CodePage = CP_ACP;
1626 }
1627
1628#if wxUSE_FONTMAP
1629 wxMBConv_win32(const wxChar* name)
1630 {
1631 m_CodePage = wxCharsetToCodepage(name);
1632 }
1633
1634 wxMBConv_win32(wxFontEncoding encoding)
1635 {
1636 m_CodePage = wxEncodingToCodepage(encoding);
1637 }
1638#endif
1639
1640 size_t MB2WC(wchar_t *buf, const char *psz, size_t n) const
1641 {
1642 // note that we have to use MB_ERR_INVALID_CHARS flag as it without it
1643 // the behaviour is not compatible with the Unix version (using iconv)
1644 // and break the library itself, e.g. wxTextInputStream::NextChar()
1645 // wouldn't work if reading an incomplete MB char didn't result in an
1646 // error
1647 //
1648 // note however that using MB_ERR_INVALID_CHARS with CP_UTF7 results in
1649 // an error (tested under Windows Server 2003) and apparently it is
1650 // done on purpose, i.e. the function accepts any input in this case
1651 // and although I'd prefer to return error on ill-formed output, our
1652 // own wxMBConvUTF7 doesn't detect errors (e.g. lone "+" which is
1653 // explicitly ill-formed according to RFC 2152) neither so we don't
1654 // even have any fallback here...
1655 int flags = m_CodePage == CP_UTF7 ? 0 : MB_ERR_INVALID_CHARS;
1656
1657 const size_t len = ::MultiByteToWideChar
1658 (
1659 m_CodePage, // code page
1660 flags, // flags: fall on error
1661 psz, // input string
1662 -1, // its length (NUL-terminated)
1663 buf, // output string
1664 buf ? n : 0 // size of output buffer
1665 );
1666
1667 // note that it returns count of written chars for buf != NULL and size
1668 // of the needed buffer for buf == NULL so in either case the length of
1669 // the string (which never includes the terminating NUL) is one less
1670 return len ? len - 1 : (size_t)-1;
1671 }
1672
1673 size_t WC2MB(char *buf, const wchar_t *pwz, size_t n) const
1674 {
1675 /*
1676 we have a problem here: by default, WideCharToMultiByte() may
1677 replace characters unrepresentable in the target code page with bad
1678 quality approximations such as turning "1/2" symbol (U+00BD) into
1679 "1" for the code pages which don't have it and we, obviously, want
1680 to avoid this at any price
1681
1682 the trouble is that this function does it _silently_, i.e. it won't
1683 even tell us whether it did or not... Win98/2000 and higher provide
1684 WC_NO_BEST_FIT_CHARS but it doesn't work for the older systems and
1685 we have to resort to a round trip, i.e. check that converting back
1686 results in the same string -- this is, of course, expensive but
1687 otherwise we simply can't be sure to not garble the data.
1688 */
1689
1690 // determine if we can rely on WC_NO_BEST_FIT_CHARS: according to MSDN
1691 // it doesn't work with CJK encodings (which we test for rather roughly
1692 // here...) nor with UTF-7/8 nor, of course, with Windows versions not
1693 // supporting it
1694 BOOL usedDef wxDUMMY_INITIALIZE(false);
1695 BOOL *pUsedDef;
1696 int flags;
1697 if ( CanUseNoBestFit() && m_CodePage < 50000 )
1698 {
1699 // it's our lucky day
1700 flags = WC_NO_BEST_FIT_CHARS;
1701 pUsedDef = &usedDef;
1702 }
1703 else // old system or unsupported encoding
1704 {
1705 flags = 0;
1706 pUsedDef = NULL;
1707 }
1708
1709 const size_t len = ::WideCharToMultiByte
1710 (
1711 m_CodePage, // code page
1712 flags, // either none or no best fit
1713 pwz, // input string
1714 -1, // it is (wide) NUL-terminated
1715 buf, // output buffer
1716 buf ? n : 0, // and its size
1717 NULL, // default "replacement" char
1718 pUsedDef // [out] was it used?
1719 );
1720
1721 if ( !len )
1722 {
1723 // function totally failed
1724 return (size_t)-1;
1725 }
1726
1727 // if we were really converting, check if we succeeded
1728 if ( buf )
1729 {
1730 if ( flags )
1731 {
1732 // check if the conversion failed, i.e. if any replacements
1733 // were done
1734 if ( usedDef )
1735 return (size_t)-1;
1736 }
1737 else // we must resort to double tripping...
1738 {
1739 wxWCharBuffer wcBuf(n);
1740 if ( MB2WC(wcBuf.data(), buf, n) == (size_t)-1 ||
1741 wcscmp(wcBuf, pwz) != 0 )
1742 {
1743 // we didn't obtain the same thing we started from, hence
1744 // the conversion was lossy and we consider that it failed
1745 return (size_t)-1;
1746 }
1747 }
1748 }
1749
1750 // see the comment above for the reason of "len - 1"
1751 return len - 1;
1752 }
1753
1754 bool IsOk() const { return m_CodePage != -1; }
1755
1756private:
1757 static bool CanUseNoBestFit()
1758 {
1759 static int s_isWin98Or2k = -1;
1760
1761 if ( s_isWin98Or2k == -1 )
1762 {
1763 int verMaj, verMin;
1764 switch ( wxGetOsVersion(&verMaj, &verMin) )
1765 {
1766 case wxWIN95:
1767 s_isWin98Or2k = verMaj >= 4 && verMin >= 10;
1768 break;
1769
1770 case wxWINDOWS_NT:
1771 s_isWin98Or2k = verMaj >= 5;
1772 break;
1773
1774 default:
1775 // unknown, be conseravtive by default
1776 s_isWin98Or2k = 0;
1777 }
1778
1779 wxASSERT_MSG( s_isWin98Or2k != -1, _T("should be set above") );
1780 }
1781
1782 return s_isWin98Or2k == 1;
1783 }
1784
1785 long m_CodePage;
1786};
1787
1788#endif // wxHAVE_WIN32_MB2WC
1789
1790// ============================================================================
1791// Cocoa conversion classes
1792// ============================================================================
1793
1794#if defined(__WXCOCOA__)
1795
1796// RN: There is no UTF-32 support in either Core Foundation or
1797// Cocoa. Strangely enough, internally Core Foundation uses
1798// UTF 32 internally quite a bit - its just not public (yet).
1799
1800#include <CoreFoundation/CFString.h>
1801#include <CoreFoundation/CFStringEncodingExt.h>
1802
1803CFStringEncoding wxCFStringEncFromFontEnc(wxFontEncoding encoding)
1804{
1805 CFStringEncoding enc = kCFStringEncodingInvalidId ;
1806 if ( encoding == wxFONTENCODING_DEFAULT )
1807 {
1808 enc = CFStringGetSystemEncoding();
1809 }
1810 else switch( encoding)
1811 {
1812 case wxFONTENCODING_ISO8859_1 :
1813 enc = kCFStringEncodingISOLatin1 ;
1814 break ;
1815 case wxFONTENCODING_ISO8859_2 :
1816 enc = kCFStringEncodingISOLatin2;
1817 break ;
1818 case wxFONTENCODING_ISO8859_3 :
1819 enc = kCFStringEncodingISOLatin3 ;
1820 break ;
1821 case wxFONTENCODING_ISO8859_4 :
1822 enc = kCFStringEncodingISOLatin4;
1823 break ;
1824 case wxFONTENCODING_ISO8859_5 :
1825 enc = kCFStringEncodingISOLatinCyrillic;
1826 break ;
1827 case wxFONTENCODING_ISO8859_6 :
1828 enc = kCFStringEncodingISOLatinArabic;
1829 break ;
1830 case wxFONTENCODING_ISO8859_7 :
1831 enc = kCFStringEncodingISOLatinGreek;
1832 break ;
1833 case wxFONTENCODING_ISO8859_8 :
1834 enc = kCFStringEncodingISOLatinHebrew;
1835 break ;
1836 case wxFONTENCODING_ISO8859_9 :
1837 enc = kCFStringEncodingISOLatin5;
1838 break ;
1839 case wxFONTENCODING_ISO8859_10 :
1840 enc = kCFStringEncodingISOLatin6;
1841 break ;
1842 case wxFONTENCODING_ISO8859_11 :
1843 enc = kCFStringEncodingISOLatinThai;
1844 break ;
1845 case wxFONTENCODING_ISO8859_13 :
1846 enc = kCFStringEncodingISOLatin7;
1847 break ;
1848 case wxFONTENCODING_ISO8859_14 :
1849 enc = kCFStringEncodingISOLatin8;
1850 break ;
1851 case wxFONTENCODING_ISO8859_15 :
1852 enc = kCFStringEncodingISOLatin9;
1853 break ;
1854
1855 case wxFONTENCODING_KOI8 :
1856 enc = kCFStringEncodingKOI8_R;
1857 break ;
1858 case wxFONTENCODING_ALTERNATIVE : // MS-DOS CP866
1859 enc = kCFStringEncodingDOSRussian;
1860 break ;
1861
1862// case wxFONTENCODING_BULGARIAN :
1863// enc = ;
1864// break ;
1865
1866 case wxFONTENCODING_CP437 :
1867 enc =kCFStringEncodingDOSLatinUS ;
1868 break ;
1869 case wxFONTENCODING_CP850 :
1870 enc = kCFStringEncodingDOSLatin1;
1871 break ;
1872 case wxFONTENCODING_CP852 :
1873 enc = kCFStringEncodingDOSLatin2;
1874 break ;
1875 case wxFONTENCODING_CP855 :
1876 enc = kCFStringEncodingDOSCyrillic;
1877 break ;
1878 case wxFONTENCODING_CP866 :
1879 enc =kCFStringEncodingDOSRussian ;
1880 break ;
1881 case wxFONTENCODING_CP874 :
1882 enc = kCFStringEncodingDOSThai;
1883 break ;
1884 case wxFONTENCODING_CP932 :
1885 enc = kCFStringEncodingDOSJapanese;
1886 break ;
1887 case wxFONTENCODING_CP936 :
1888 enc =kCFStringEncodingDOSChineseSimplif ;
1889 break ;
1890 case wxFONTENCODING_CP949 :
1891 enc = kCFStringEncodingDOSKorean;
1892 break ;
1893 case wxFONTENCODING_CP950 :
1894 enc = kCFStringEncodingDOSChineseTrad;
1895 break ;
1896 case wxFONTENCODING_CP1250 :
1897 enc = kCFStringEncodingWindowsLatin2;
1898 break ;
1899 case wxFONTENCODING_CP1251 :
1900 enc =kCFStringEncodingWindowsCyrillic ;
1901 break ;
1902 case wxFONTENCODING_CP1252 :
1903 enc =kCFStringEncodingWindowsLatin1 ;
1904 break ;
1905 case wxFONTENCODING_CP1253 :
1906 enc = kCFStringEncodingWindowsGreek;
1907 break ;
1908 case wxFONTENCODING_CP1254 :
1909 enc = kCFStringEncodingWindowsLatin5;
1910 break ;
1911 case wxFONTENCODING_CP1255 :
1912 enc =kCFStringEncodingWindowsHebrew ;
1913 break ;
1914 case wxFONTENCODING_CP1256 :
1915 enc =kCFStringEncodingWindowsArabic ;
1916 break ;
1917 case wxFONTENCODING_CP1257 :
1918 enc = kCFStringEncodingWindowsBalticRim;
1919 break ;
1920// This only really encodes to UTF7 (if that) evidently
1921// case wxFONTENCODING_UTF7 :
1922// enc = kCFStringEncodingNonLossyASCII ;
1923// break ;
1924 case wxFONTENCODING_UTF8 :
1925 enc = kCFStringEncodingUTF8 ;
1926 break ;
1927 case wxFONTENCODING_EUC_JP :
1928 enc = kCFStringEncodingEUC_JP;
1929 break ;
1930 case wxFONTENCODING_UTF16 :
1931 enc = kCFStringEncodingUnicode ;
1932 break ;
1933 case wxFONTENCODING_MACROMAN :
1934 enc = kCFStringEncodingMacRoman ;
1935 break ;
1936 case wxFONTENCODING_MACJAPANESE :
1937 enc = kCFStringEncodingMacJapanese ;
1938 break ;
1939 case wxFONTENCODING_MACCHINESETRAD :
1940 enc = kCFStringEncodingMacChineseTrad ;
1941 break ;
1942 case wxFONTENCODING_MACKOREAN :
1943 enc = kCFStringEncodingMacKorean ;
1944 break ;
1945 case wxFONTENCODING_MACARABIC :
1946 enc = kCFStringEncodingMacArabic ;
1947 break ;
1948 case wxFONTENCODING_MACHEBREW :
1949 enc = kCFStringEncodingMacHebrew ;
1950 break ;
1951 case wxFONTENCODING_MACGREEK :
1952 enc = kCFStringEncodingMacGreek ;
1953 break ;
1954 case wxFONTENCODING_MACCYRILLIC :
1955 enc = kCFStringEncodingMacCyrillic ;
1956 break ;
1957 case wxFONTENCODING_MACDEVANAGARI :
1958 enc = kCFStringEncodingMacDevanagari ;
1959 break ;
1960 case wxFONTENCODING_MACGURMUKHI :
1961 enc = kCFStringEncodingMacGurmukhi ;
1962 break ;
1963 case wxFONTENCODING_MACGUJARATI :
1964 enc = kCFStringEncodingMacGujarati ;
1965 break ;
1966 case wxFONTENCODING_MACORIYA :
1967 enc = kCFStringEncodingMacOriya ;
1968 break ;
1969 case wxFONTENCODING_MACBENGALI :
1970 enc = kCFStringEncodingMacBengali ;
1971 break ;
1972 case wxFONTENCODING_MACTAMIL :
1973 enc = kCFStringEncodingMacTamil ;
1974 break ;
1975 case wxFONTENCODING_MACTELUGU :
1976 enc = kCFStringEncodingMacTelugu ;
1977 break ;
1978 case wxFONTENCODING_MACKANNADA :
1979 enc = kCFStringEncodingMacKannada ;
1980 break ;
1981 case wxFONTENCODING_MACMALAJALAM :
1982 enc = kCFStringEncodingMacMalayalam ;
1983 break ;
1984 case wxFONTENCODING_MACSINHALESE :
1985 enc = kCFStringEncodingMacSinhalese ;
1986 break ;
1987 case wxFONTENCODING_MACBURMESE :
1988 enc = kCFStringEncodingMacBurmese ;
1989 break ;
1990 case wxFONTENCODING_MACKHMER :
1991 enc = kCFStringEncodingMacKhmer ;
1992 break ;
1993 case wxFONTENCODING_MACTHAI :
1994 enc = kCFStringEncodingMacThai ;
1995 break ;
1996 case wxFONTENCODING_MACLAOTIAN :
1997 enc = kCFStringEncodingMacLaotian ;
1998 break ;
1999 case wxFONTENCODING_MACGEORGIAN :
2000 enc = kCFStringEncodingMacGeorgian ;
2001 break ;
2002 case wxFONTENCODING_MACARMENIAN :
2003 enc = kCFStringEncodingMacArmenian ;
2004 break ;
2005 case wxFONTENCODING_MACCHINESESIMP :
2006 enc = kCFStringEncodingMacChineseSimp ;
2007 break ;
2008 case wxFONTENCODING_MACTIBETAN :
2009 enc = kCFStringEncodingMacTibetan ;
2010 break ;
2011 case wxFONTENCODING_MACMONGOLIAN :
2012 enc = kCFStringEncodingMacMongolian ;
2013 break ;
2014 case wxFONTENCODING_MACETHIOPIC :
2015 enc = kCFStringEncodingMacEthiopic ;
2016 break ;
2017 case wxFONTENCODING_MACCENTRALEUR :
2018 enc = kCFStringEncodingMacCentralEurRoman ;
2019 break ;
2020 case wxFONTENCODING_MACVIATNAMESE :
2021 enc = kCFStringEncodingMacVietnamese ;
2022 break ;
2023 case wxFONTENCODING_MACARABICEXT :
2024 enc = kCFStringEncodingMacExtArabic ;
2025 break ;
2026 case wxFONTENCODING_MACSYMBOL :
2027 enc = kCFStringEncodingMacSymbol ;
2028 break ;
2029 case wxFONTENCODING_MACDINGBATS :
2030 enc = kCFStringEncodingMacDingbats ;
2031 break ;
2032 case wxFONTENCODING_MACTURKISH :
2033 enc = kCFStringEncodingMacTurkish ;
2034 break ;
2035 case wxFONTENCODING_MACCROATIAN :
2036 enc = kCFStringEncodingMacCroatian ;
2037 break ;
2038 case wxFONTENCODING_MACICELANDIC :
2039 enc = kCFStringEncodingMacIcelandic ;
2040 break ;
2041 case wxFONTENCODING_MACROMANIAN :
2042 enc = kCFStringEncodingMacRomanian ;
2043 break ;
2044 case wxFONTENCODING_MACCELTIC :
2045 enc = kCFStringEncodingMacCeltic ;
2046 break ;
2047 case wxFONTENCODING_MACGAELIC :
2048 enc = kCFStringEncodingMacGaelic ;
2049 break ;
2050// case wxFONTENCODING_MACKEYBOARD :
2051// enc = kCFStringEncodingMacKeyboardGlyphs ;
2052// break ;
2053 default :
2054 // because gcc is picky
2055 break ;
2056 } ;
2057 return enc ;
2058}
2059
2060class wxMBConv_cocoa : public wxMBConv
2061{
2062public:
2063 wxMBConv_cocoa()
2064 {
2065 Init(CFStringGetSystemEncoding()) ;
2066 }
2067
2068#if wxUSE_FONTMAP
2069 wxMBConv_cocoa(const wxChar* name)
2070 {
2071 Init( wxCFStringEncFromFontEnc(wxFontMapperBase::Get()->CharsetToEncoding(name, false) ) ) ;
2072 }
2073#endif
2074
2075 wxMBConv_cocoa(wxFontEncoding encoding)
2076 {
2077 Init( wxCFStringEncFromFontEnc(encoding) );
2078 }
2079
2080 ~wxMBConv_cocoa()
2081 {
2082 }
2083
2084 void Init( CFStringEncoding encoding)
2085 {
2086 m_encoding = encoding ;
2087 }
2088
2089 size_t MB2WC(wchar_t * szOut, const char * szUnConv, size_t nOutSize) const
2090 {
2091 wxASSERT(szUnConv);
2092
2093 CFStringRef theString = CFStringCreateWithBytes (
2094 NULL, //the allocator
2095 (const UInt8*)szUnConv,
2096 strlen(szUnConv),
2097 m_encoding,
2098 false //no BOM/external representation
2099 );
2100
2101 wxASSERT(theString);
2102
2103 size_t nOutLength = CFStringGetLength(theString);
2104
2105 if (szOut == NULL)
2106 {
2107 CFRelease(theString);
2108 return nOutLength;
2109 }
2110
2111 CFRange theRange = { 0, nOutSize };
2112
2113#if SIZEOF_WCHAR_T == 4
2114 UniChar* szUniCharBuffer = new UniChar[nOutSize];
2115#endif
2116
2117 CFStringGetCharacters(theString, theRange, szUniCharBuffer);
2118
2119 CFRelease(theString);
2120
2121 szUniCharBuffer[nOutLength] = '\0' ;
2122
2123#if SIZEOF_WCHAR_T == 4
2124 wxMBConvUTF16 converter ;
2125 converter.MB2WC(szOut, (const char*)szUniCharBuffer , nOutSize ) ;
2126 delete[] szUniCharBuffer;
2127#endif
2128
2129 return nOutLength;
2130 }
2131
2132 size_t WC2MB(char *szOut, const wchar_t *szUnConv, size_t nOutSize) const
2133 {
2134 wxASSERT(szUnConv);
2135
2136 size_t nRealOutSize;
2137 size_t nBufSize = wxWcslen(szUnConv);
2138 UniChar* szUniBuffer = (UniChar*) szUnConv;
2139
2140#if SIZEOF_WCHAR_T == 4
2141 wxMBConvUTF16 converter ;
2142 nBufSize = converter.WC2MB( NULL , szUnConv , 0 );
2143 szUniBuffer = new UniChar[ (nBufSize / sizeof(UniChar)) + 1] ;
2144 converter.WC2MB( (char*) szUniBuffer , szUnConv, nBufSize + sizeof(UniChar)) ;
2145 nBufSize /= sizeof(UniChar);
2146#endif
2147
2148 CFStringRef theString = CFStringCreateWithCharactersNoCopy(
2149 NULL, //allocator
2150 szUniBuffer,
2151 nBufSize,
2152 kCFAllocatorNull //deallocator - we want to deallocate it ourselves
2153 );
2154
2155 wxASSERT(theString);
2156
2157 //Note that CER puts a BOM when converting to unicode
2158 //so we check and use getchars instead in that case
2159 if (m_encoding == kCFStringEncodingUnicode)
2160 {
2161 if (szOut != NULL)
2162 CFStringGetCharacters(theString, CFRangeMake(0, nOutSize - 1), (UniChar*) szOut);
2163
2164 nRealOutSize = CFStringGetLength(theString) + 1;
2165 }
2166 else
2167 {
2168 CFStringGetBytes(
2169 theString,
2170 CFRangeMake(0, CFStringGetLength(theString)),
2171 m_encoding,
2172 0, //what to put in characters that can't be converted -
2173 //0 tells CFString to return NULL if it meets such a character
2174 false, //not an external representation
2175 (UInt8*) szOut,
2176 nOutSize,
2177 (CFIndex*) &nRealOutSize
2178 );
2179 }
2180
2181 CFRelease(theString);
2182
2183#if SIZEOF_WCHAR_T == 4
2184 delete[] szUniBuffer;
2185#endif
2186
2187 return nRealOutSize - 1;
2188 }
2189
2190 bool IsOk() const
2191 {
2192 return m_encoding != kCFStringEncodingInvalidId &&
2193 CFStringIsEncodingAvailable(m_encoding);
2194 }
2195
2196private:
2197 CFStringEncoding m_encoding ;
2198};
2199
2200#endif // defined(__WXCOCOA__)
2201
2202// ============================================================================
2203// Mac conversion classes
2204// ============================================================================
2205
2206#if defined(__WXMAC__) && defined(TARGET_CARBON)
2207
2208class wxMBConv_mac : public wxMBConv
2209{
2210public:
2211 wxMBConv_mac()
2212 {
2213 Init(CFStringGetSystemEncoding()) ;
2214 }
2215
2216#if wxUSE_FONTMAP
2217 wxMBConv_mac(const wxChar* name)
2218 {
2219 Init( wxMacGetSystemEncFromFontEnc(wxFontMapperBase::Get()->CharsetToEncoding(name, false) ) ) ;
2220 }
2221#endif
2222
2223 wxMBConv_mac(wxFontEncoding encoding)
2224 {
2225 Init( wxMacGetSystemEncFromFontEnc(encoding) );
2226 }
2227
2228 ~wxMBConv_mac()
2229 {
2230 OSStatus status = noErr ;
2231 status = TECDisposeConverter(m_MB2WC_converter);
2232 status = TECDisposeConverter(m_WC2MB_converter);
2233 }
2234
2235
2236 void Init( TextEncodingBase encoding)
2237 {
2238 OSStatus status = noErr ;
2239 m_char_encoding = encoding ;
2240 m_unicode_encoding = CreateTextEncoding(kTextEncodingUnicodeDefault,0,kUnicode16BitFormat) ;
2241
2242 status = TECCreateConverter(&m_MB2WC_converter,
2243 m_char_encoding,
2244 m_unicode_encoding);
2245 status = TECCreateConverter(&m_WC2MB_converter,
2246 m_unicode_encoding,
2247 m_char_encoding);
2248 }
2249
2250 size_t MB2WC(wchar_t *buf, const char *psz, size_t n) const
2251 {
2252 OSStatus status = noErr ;
2253 ByteCount byteOutLen ;
2254 ByteCount byteInLen = strlen(psz) ;
2255 wchar_t *tbuf = NULL ;
2256 UniChar* ubuf = NULL ;
2257 size_t res = 0 ;
2258
2259 if (buf == NULL)
2260 {
2261 //apple specs say at least 32
2262 n = wxMax( 32 , byteInLen ) ;
2263 tbuf = (wchar_t*) malloc( n * SIZEOF_WCHAR_T) ;
2264 }
2265 ByteCount byteBufferLen = n * sizeof( UniChar ) ;
2266#if SIZEOF_WCHAR_T == 4
2267 ubuf = (UniChar*) malloc( byteBufferLen + 2 ) ;
2268#else
2269 ubuf = (UniChar*) (buf ? buf : tbuf) ;
2270#endif
2271 status = TECConvertText(m_MB2WC_converter, (ConstTextPtr) psz , byteInLen, &byteInLen,
2272 (TextPtr) ubuf , byteBufferLen, &byteOutLen);
2273#if SIZEOF_WCHAR_T == 4
2274 // we have to terminate here, because n might be larger for the trailing zero, and if UniChar
2275 // is not properly terminated we get random characters at the end
2276 ubuf[byteOutLen / sizeof( UniChar ) ] = 0 ;
2277 wxMBConvUTF16 converter ;
2278 res = converter.MB2WC( (buf ? buf : tbuf) , (const char*)ubuf , n ) ;
2279 free( ubuf ) ;
2280#else
2281 res = byteOutLen / sizeof( UniChar ) ;
2282#endif
2283 if ( buf == NULL )
2284 free(tbuf) ;
2285
2286 if ( buf && res < n)
2287 buf[res] = 0;
2288
2289 return res ;
2290 }
2291
2292 size_t WC2MB(char *buf, const wchar_t *psz, size_t n) const
2293 {
2294 OSStatus status = noErr ;
2295 ByteCount byteOutLen ;
2296 ByteCount byteInLen = wxWcslen(psz) * SIZEOF_WCHAR_T ;
2297
2298 char *tbuf = NULL ;
2299
2300 if (buf == NULL)
2301 {
2302 //apple specs say at least 32
2303 n = wxMax( 32 , ((byteInLen / SIZEOF_WCHAR_T) * 8) + SIZEOF_WCHAR_T );
2304 tbuf = (char*) malloc( n ) ;
2305 }
2306
2307 ByteCount byteBufferLen = n ;
2308 UniChar* ubuf = NULL ;
2309#if SIZEOF_WCHAR_T == 4
2310 wxMBConvUTF16 converter ;
2311 size_t unicharlen = converter.WC2MB( NULL , psz , 0 ) ;
2312 byteInLen = unicharlen ;
2313 ubuf = (UniChar*) malloc( byteInLen + 2 ) ;
2314 converter.WC2MB( (char*) ubuf , psz, unicharlen + 2 ) ;
2315#else
2316 ubuf = (UniChar*) psz ;
2317#endif
2318 status = TECConvertText(m_WC2MB_converter, (ConstTextPtr) ubuf , byteInLen, &byteInLen,
2319 (TextPtr) (buf ? buf : tbuf) , byteBufferLen, &byteOutLen);
2320#if SIZEOF_WCHAR_T == 4
2321 free( ubuf ) ;
2322#endif
2323 if ( buf == NULL )
2324 free(tbuf) ;
2325
2326 size_t res = byteOutLen ;
2327 if ( buf && res < n)
2328 {
2329 buf[res] = 0;
2330
2331 //we need to double-trip to verify it didn't insert any ? in place
2332 //of bogus characters
2333 wxWCharBuffer wcBuf(n);
2334 size_t pszlen = wxWcslen(psz);
2335 if ( MB2WC(wcBuf.data(), buf, n) == (size_t)-1 ||
2336 wxWcslen(wcBuf) != pszlen ||
2337 memcmp(wcBuf, psz, pszlen * sizeof(wchar_t)) != 0 )
2338 {
2339 // we didn't obtain the same thing we started from, hence
2340 // the conversion was lossy and we consider that it failed
2341 return (size_t)-1;
2342 }
2343 }
2344
2345 return res ;
2346 }
2347
2348 bool IsOk() const
2349 { return m_MB2WC_converter != NULL && m_WC2MB_converter != NULL ; }
2350
2351private:
2352 TECObjectRef m_MB2WC_converter ;
2353 TECObjectRef m_WC2MB_converter ;
2354
2355 TextEncodingBase m_char_encoding ;
2356 TextEncodingBase m_unicode_encoding ;
2357};
2358
2359#endif // defined(__WXMAC__) && defined(TARGET_CARBON)
2360
2361// ============================================================================
2362// wxEncodingConverter based conversion classes
2363// ============================================================================
2364
2365#if wxUSE_FONTMAP
2366
2367class wxMBConv_wxwin : public wxMBConv
2368{
2369private:
2370 void Init()
2371 {
2372 m_ok = m2w.Init(m_enc, wxFONTENCODING_UNICODE) &&
2373 w2m.Init(wxFONTENCODING_UNICODE, m_enc);
2374 }
2375
2376public:
2377 // temporarily just use wxEncodingConverter stuff,
2378 // so that it works while a better implementation is built
2379 wxMBConv_wxwin(const wxChar* name)
2380 {
2381 if (name)
2382 m_enc = wxFontMapperBase::Get()->CharsetToEncoding(name, false);
2383 else
2384 m_enc = wxFONTENCODING_SYSTEM;
2385
2386 Init();
2387 }
2388
2389 wxMBConv_wxwin(wxFontEncoding enc)
2390 {
2391 m_enc = enc;
2392
2393 Init();
2394 }
2395
2396 size_t MB2WC(wchar_t *buf, const char *psz, size_t WXUNUSED(n)) const
2397 {
2398 size_t inbuf = strlen(psz);
2399 if (buf)
2400 {
2401 if (!m2w.Convert(psz,buf))
2402 return (size_t)-1;
2403 }
2404 return inbuf;
2405 }
2406
2407 size_t WC2MB(char *buf, const wchar_t *psz, size_t WXUNUSED(n)) const
2408 {
2409 const size_t inbuf = wxWcslen(psz);
2410 if (buf)
2411 {
2412 if (!w2m.Convert(psz,buf))
2413 return (size_t)-1;
2414 }
2415
2416 return inbuf;
2417 }
2418
2419 bool IsOk() const { return m_ok; }
2420
2421public:
2422 wxFontEncoding m_enc;
2423 wxEncodingConverter m2w, w2m;
2424
2425 // were we initialized successfully?
2426 bool m_ok;
2427
2428 DECLARE_NO_COPY_CLASS(wxMBConv_wxwin)
2429};
2430
2431// make the constructors available for unit testing
2432WXDLLIMPEXP_BASE wxMBConv* new_wxMBConv_wxwin( const wxChar* name )
2433{
2434 wxMBConv_wxwin* result = new wxMBConv_wxwin( name );
2435 if ( !result->IsOk() )
2436 {
2437 delete result;
2438 return 0;
2439 }
2440 return result;
2441}
2442
2443#endif // wxUSE_FONTMAP
2444
2445// ============================================================================
2446// wxCSConv implementation
2447// ============================================================================
2448
2449void wxCSConv::Init()
2450{
2451 m_name = NULL;
2452 m_convReal = NULL;
2453 m_deferred = true;
2454}
2455
2456wxCSConv::wxCSConv(const wxChar *charset)
2457{
2458 Init();
2459
2460 if ( charset )
2461 {
2462 SetName(charset);
2463 }
2464
2465#if wxUSE_FONTMAP
2466 m_encoding = wxFontMapperBase::GetEncodingFromName(charset);
2467#else
2468 m_encoding = wxFONTENCODING_SYSTEM;
2469#endif
2470}
2471
2472wxCSConv::wxCSConv(wxFontEncoding encoding)
2473{
2474 if ( encoding == wxFONTENCODING_MAX || encoding == wxFONTENCODING_DEFAULT )
2475 {
2476 wxFAIL_MSG( _T("invalid encoding value in wxCSConv ctor") );
2477
2478 encoding = wxFONTENCODING_SYSTEM;
2479 }
2480
2481 Init();
2482
2483 m_encoding = encoding;
2484}
2485
2486wxCSConv::~wxCSConv()
2487{
2488 Clear();
2489}
2490
2491wxCSConv::wxCSConv(const wxCSConv& conv)
2492 : wxMBConv()
2493{
2494 Init();
2495
2496 SetName(conv.m_name);
2497 m_encoding = conv.m_encoding;
2498}
2499
2500wxCSConv& wxCSConv::operator=(const wxCSConv& conv)
2501{
2502 Clear();
2503
2504 SetName(conv.m_name);
2505 m_encoding = conv.m_encoding;
2506
2507 return *this;
2508}
2509
2510void wxCSConv::Clear()
2511{
2512 free(m_name);
2513 delete m_convReal;
2514
2515 m_name = NULL;
2516 m_convReal = NULL;
2517}
2518
2519void wxCSConv::SetName(const wxChar *charset)
2520{
2521 if (charset)
2522 {
2523 m_name = wxStrdup(charset);
2524 m_deferred = true;
2525 }
2526}
2527
2528#if wxUSE_FONTMAP
2529#include "wx/hashmap.h"
2530
2531WX_DECLARE_HASH_MAP( wxFontEncoding, wxString, wxIntegerHash, wxIntegerEqual,
2532 wxEncodingNameCache );
2533
2534static wxEncodingNameCache gs_nameCache;
2535#endif
2536
2537wxMBConv *wxCSConv::DoCreate() const
2538{
2539#if wxUSE_FONTMAP
2540 wxLogTrace(TRACE_STRCONV,
2541 wxT("creating conversion for %s"),
2542 (m_name ? m_name
2543 : wxFontMapperBase::GetEncodingName(m_encoding).c_str()));
2544#endif // wxUSE_FONTMAP
2545
2546 // check for the special case of ASCII or ISO8859-1 charset: as we have
2547 // special knowledge of it anyhow, we don't need to create a special
2548 // conversion object
2549 if ( m_encoding == wxFONTENCODING_ISO8859_1 ||
2550 m_encoding == wxFONTENCODING_DEFAULT )
2551 {
2552 // don't convert at all
2553 return NULL;
2554 }
2555
2556 // we trust OS to do conversion better than we can so try external
2557 // conversion methods first
2558 //
2559 // the full order is:
2560 // 1. OS conversion (iconv() under Unix or Win32 API)
2561 // 2. hard coded conversions for UTF
2562 // 3. wxEncodingConverter as fall back
2563
2564 // step (1)
2565#ifdef HAVE_ICONV
2566#if !wxUSE_FONTMAP
2567 if ( m_name )
2568#endif // !wxUSE_FONTMAP
2569 {
2570 wxString name(m_name);
2571 wxFontEncoding encoding(m_encoding);
2572
2573 if ( !name.empty() )
2574 {
2575 wxMBConv_iconv *conv = new wxMBConv_iconv(name);
2576 if ( conv->IsOk() )
2577 return conv;
2578
2579 delete conv;
2580
2581#if wxUSE_FONTMAP
2582 encoding =
2583 wxFontMapperBase::Get()->CharsetToEncoding(name, false);
2584#endif // wxUSE_FONTMAP
2585 }
2586#if wxUSE_FONTMAP
2587 {
2588 const wxEncodingNameCache::iterator it = gs_nameCache.find(encoding);
2589 if ( it != gs_nameCache.end() )
2590 {
2591 if ( it->second.empty() )
2592 return NULL;
2593
2594 wxMBConv_iconv *conv = new wxMBConv_iconv(it->second);
2595 if ( conv->IsOk() )
2596 return conv;
2597
2598 delete conv;
2599 }
2600
2601 const wxChar** names = wxFontMapperBase::GetAllEncodingNames(encoding);
2602
2603 for ( ; *names; ++names )
2604 {
2605 wxMBConv_iconv *conv = new wxMBConv_iconv(*names);
2606 if ( conv->IsOk() )
2607 {
2608 gs_nameCache[encoding] = *names;
2609 return conv;
2610 }
2611
2612 delete conv;
2613 }
2614
2615 gs_nameCache[encoding] = _T(""); // cache the failure
2616 }
2617#endif // wxUSE_FONTMAP
2618 }
2619#endif // HAVE_ICONV
2620
2621#ifdef wxHAVE_WIN32_MB2WC
2622 {
2623#if wxUSE_FONTMAP
2624 wxMBConv_win32 *conv = m_name ? new wxMBConv_win32(m_name)
2625 : new wxMBConv_win32(m_encoding);
2626 if ( conv->IsOk() )
2627 return conv;
2628
2629 delete conv;
2630#else
2631 return NULL;
2632#endif
2633 }
2634#endif // wxHAVE_WIN32_MB2WC
2635#if defined(__WXMAC__)
2636 {
2637 // leave UTF16 and UTF32 to the built-ins of wx
2638 if ( m_name || ( m_encoding < wxFONTENCODING_UTF16BE ||
2639 ( m_encoding >= wxFONTENCODING_MACMIN && m_encoding <= wxFONTENCODING_MACMAX ) ) )
2640 {
2641
2642#if wxUSE_FONTMAP
2643 wxMBConv_mac *conv = m_name ? new wxMBConv_mac(m_name)
2644 : new wxMBConv_mac(m_encoding);
2645#else
2646 wxMBConv_mac *conv = new wxMBConv_mac(m_encoding);
2647#endif
2648 if ( conv->IsOk() )
2649 return conv;
2650
2651 delete conv;
2652 }
2653 }
2654#endif
2655#if defined(__WXCOCOA__)
2656 {
2657 if ( m_name || ( m_encoding <= wxFONTENCODING_UTF16 ) )
2658 {
2659
2660#if wxUSE_FONTMAP
2661 wxMBConv_cocoa *conv = m_name ? new wxMBConv_cocoa(m_name)
2662 : new wxMBConv_cocoa(m_encoding);
2663#else
2664 wxMBConv_cocoa *conv = new wxMBConv_cocoa(m_encoding);
2665#endif
2666 if ( conv->IsOk() )
2667 return conv;
2668
2669 delete conv;
2670 }
2671 }
2672#endif
2673 // step (2)
2674 wxFontEncoding enc = m_encoding;
2675#if wxUSE_FONTMAP
2676 if ( enc == wxFONTENCODING_SYSTEM && m_name )
2677 {
2678 // use "false" to suppress interactive dialogs -- we can be called from
2679 // anywhere and popping up a dialog from here is the last thing we want to
2680 // do
2681 enc = wxFontMapperBase::Get()->CharsetToEncoding(m_name, false);
2682 }
2683#endif // wxUSE_FONTMAP
2684
2685 switch ( enc )
2686 {
2687 case wxFONTENCODING_UTF7:
2688 return new wxMBConvUTF7;
2689
2690 case wxFONTENCODING_UTF8:
2691 return new wxMBConvUTF8;
2692
2693 case wxFONTENCODING_UTF16BE:
2694 return new wxMBConvUTF16BE;
2695
2696 case wxFONTENCODING_UTF16LE:
2697 return new wxMBConvUTF16LE;
2698
2699 case wxFONTENCODING_UTF32BE:
2700 return new wxMBConvUTF32BE;
2701
2702 case wxFONTENCODING_UTF32LE:
2703 return new wxMBConvUTF32LE;
2704
2705 default:
2706 // nothing to do but put here to suppress gcc warnings
2707 ;
2708 }
2709
2710 // step (3)
2711#if wxUSE_FONTMAP
2712 {
2713 wxMBConv_wxwin *conv = m_name ? new wxMBConv_wxwin(m_name)
2714 : new wxMBConv_wxwin(m_encoding);
2715 if ( conv->IsOk() )
2716 return conv;
2717
2718 delete conv;
2719 }
2720#endif // wxUSE_FONTMAP
2721
2722 // NB: This is a hack to prevent deadlock. What could otherwise happen
2723 // in Unicode build: wxConvLocal creation ends up being here
2724 // because of some failure and logs the error. But wxLog will try to
2725 // attach timestamp, for which it will need wxConvLocal (to convert
2726 // time to char* and then wchar_t*), but that fails, tries to log
2727 // error, but wxLog has a (already locked) critical section that
2728 // guards static buffer.
2729 static bool alreadyLoggingError = false;
2730 if (!alreadyLoggingError)
2731 {
2732 alreadyLoggingError = true;
2733 wxLogError(_("Cannot convert from the charset '%s'!"),
2734 m_name ? m_name
2735 :
2736#if wxUSE_FONTMAP
2737 wxFontMapperBase::GetEncodingDescription(m_encoding).c_str()
2738#else // !wxUSE_FONTMAP
2739 wxString::Format(_("encoding %s"), m_encoding).c_str()
2740#endif // wxUSE_FONTMAP/!wxUSE_FONTMAP
2741 );
2742 alreadyLoggingError = false;
2743 }
2744
2745 return NULL;
2746}
2747
2748void wxCSConv::CreateConvIfNeeded() const
2749{
2750 if ( m_deferred )
2751 {
2752 wxCSConv *self = (wxCSConv *)this; // const_cast
2753
2754#if wxUSE_INTL
2755 // if we don't have neither the name nor the encoding, use the default
2756 // encoding for this system
2757 if ( !m_name && m_encoding == wxFONTENCODING_SYSTEM )
2758 {
2759 self->m_name = wxStrdup(wxLocale::GetSystemEncodingName());
2760 }
2761#endif // wxUSE_INTL
2762
2763 self->m_convReal = DoCreate();
2764 self->m_deferred = false;
2765 }
2766}
2767
2768size_t wxCSConv::MB2WC(wchar_t *buf, const char *psz, size_t n) const
2769{
2770 CreateConvIfNeeded();
2771
2772 if (m_convReal)
2773 return m_convReal->MB2WC(buf, psz, n);
2774
2775 // latin-1 (direct)
2776 size_t len = strlen(psz);
2777
2778 if (buf)
2779 {
2780 for (size_t c = 0; c <= len; c++)
2781 buf[c] = (unsigned char)(psz[c]);
2782 }
2783
2784 return len;
2785}
2786
2787size_t wxCSConv::WC2MB(char *buf, const wchar_t *psz, size_t n) const
2788{
2789 CreateConvIfNeeded();
2790
2791 if (m_convReal)
2792 return m_convReal->WC2MB(buf, psz, n);
2793
2794 // latin-1 (direct)
2795 const size_t len = wxWcslen(psz);
2796 if (buf)
2797 {
2798 for (size_t c = 0; c <= len; c++)
2799 {
2800 if (psz[c] > 0xFF)
2801 return (size_t)-1;
2802 buf[c] = (char)psz[c];
2803 }
2804 }
2805 else
2806 {
2807 for (size_t c = 0; c <= len; c++)
2808 {
2809 if (psz[c] > 0xFF)
2810 return (size_t)-1;
2811 }
2812 }
2813
2814 return len;
2815}
2816
2817// ----------------------------------------------------------------------------
2818// globals
2819// ----------------------------------------------------------------------------
2820
2821#ifdef __WINDOWS__
2822 static wxMBConv_win32 wxConvLibcObj;
2823#elif defined(__WXMAC__) && !defined(__MACH__)
2824 static wxMBConv_mac wxConvLibcObj ;
2825#else
2826 static wxMBConvLibc wxConvLibcObj;
2827#endif
2828
2829static wxCSConv wxConvLocalObj(wxFONTENCODING_SYSTEM);
2830static wxCSConv wxConvISO8859_1Obj(wxFONTENCODING_ISO8859_1);
2831static wxMBConvUTF7 wxConvUTF7Obj;
2832static wxMBConvUTF8 wxConvUTF8Obj;
2833
2834WXDLLIMPEXP_DATA_BASE(wxMBConv&) wxConvLibc = wxConvLibcObj;
2835WXDLLIMPEXP_DATA_BASE(wxCSConv&) wxConvLocal = wxConvLocalObj;
2836WXDLLIMPEXP_DATA_BASE(wxCSConv&) wxConvISO8859_1 = wxConvISO8859_1Obj;
2837WXDLLIMPEXP_DATA_BASE(wxMBConvUTF7&) wxConvUTF7 = wxConvUTF7Obj;
2838WXDLLIMPEXP_DATA_BASE(wxMBConvUTF8&) wxConvUTF8 = wxConvUTF8Obj;
2839WXDLLIMPEXP_DATA_BASE(wxMBConv *) wxConvCurrent = &wxConvLibcObj;
2840WXDLLIMPEXP_DATA_BASE(wxMBConv *) wxConvFileName = &
2841#ifdef __WXOSX__
2842 wxConvUTF8Obj;
2843#else
2844 wxConvLibcObj;
2845#endif
2846
2847
2848#else // !wxUSE_WCHAR_T
2849
2850// stand-ins in absence of wchar_t
2851WXDLLIMPEXP_DATA_BASE(wxMBConv) wxConvLibc,
2852 wxConvISO8859_1,
2853 wxConvLocal,
2854 wxConvUTF8;
2855
2856#endif // wxUSE_WCHAR_T/!wxUSE_WCHAR_T
wxWidgets (Impactor)