]> git.saurik.com Git - wxWidgets.git/blob - src/common/strconv.cpp
projects / wxWidgets.git / blob
? search:
summary | shortlog | log | commit | commitdiff | tree
blame | history | raw | HEAD
Split this out from other changes to keep things sane..
[wxWidgets.git] / src / common / strconv.cpp
1 /////////////////////////////////////////////////////////////////////////////
2 // Name: strconv.cpp
3 // Purpose: Unicode conversion classes
4 // Author: Ove Kaaven, Robert Roebling, Vadim Zeitlin, Vaclav Slavik
5 // Modified by:
6 // Created: 29/01/98
7 // RCS-ID: $Id$
8 // Copyright: (c) 1999 Ove Kaaven, Robert Roebling, Vadim Zeitlin, Vaclav Slavik
9 // Licence: wxWindows license
10 /////////////////////////////////////////////////////////////////////////////
11
12 // ============================================================================
13 // declarations
14 // ============================================================================
15
16 // ----------------------------------------------------------------------------
17 // headers
18 // ----------------------------------------------------------------------------
19
20 #ifdef __GNUG__
21 #pragma implementation "strconv.h"
22 #endif
23
24 // For compilers that support precompilation, includes "wx.h".
25 #include "wx/wxprec.h"
26
27 #ifdef __BORLANDC__
28 #pragma hdrstop
29 #endif
30
31 #ifndef WX_PRECOMP
32 #include "wx/intl.h"
33 #include "wx/log.h"
34 #endif // WX_PRECOMP
35
36 #ifdef __WXMSW__
37 #include "wx/msw/private.h"
38 #endif
39
40 #include <errno.h>
41 #include <ctype.h>
42 #include <string.h>
43 #include <stdlib.h>
44
45 #include "wx/module.h"
46 #include "wx/strconv.h"
47
48 // ----------------------------------------------------------------------------
49 // globals
50 // ----------------------------------------------------------------------------
51
52 #if wxUSE_WCHAR_T
53 WXDLLEXPORT_DATA(wxMBConv) wxConvLibc;
54 WXDLLEXPORT_DATA(wxCSConv) wxConvLocal((const wxChar *)NULL);
55 WXDLLEXPORT_DATA(wxCSConv) wxConvISO8859_1(_T("iso-8859-1"));
56 #else
57 // stand-ins in absence of wchar_t
58 WXDLLEXPORT_DATA(wxMBConv) wxConvLibc,
59 wxConvFile,
60 wxConvISO8859_1,
61 wxConvLocal;
62 #endif // wxUSE_WCHAR_T
63
64 WXDLLEXPORT_DATA(wxMBConv *) wxConvCurrent = &wxConvLibc;
65
66 class wxStrConvModule: public wxModule
67 {
68 public:
69 wxStrConvModule() : wxModule() { }
70 virtual bool OnInit() { return TRUE; }
71 virtual void OnExit()
72 {
73 #if wxUSE_WCHAR_T
74 wxConvLocal.Clear();
75 wxConvISO8859_1.Clear();
76 #endif
77 }
78
79 DECLARE_DYNAMIC_CLASS(wxStrConvModule)
80 };
81
82 IMPLEMENT_DYNAMIC_CLASS(wxStrConvModule, wxModule)
83
84
85 // ----------------------------------------------------------------------------
86 // headers
87 // ----------------------------------------------------------------------------
88
89 #if wxUSE_WCHAR_T
90
91 #ifdef __SALFORDC__
92 #include <clib.h>
93 #endif
94
95 #ifdef HAVE_ICONV
96 #include <iconv.h>
97 #endif
98
99 #include "wx/encconv.h"
100 #include "wx/fontmap.h"
101
102 // ----------------------------------------------------------------------------
103 // macros
104 // ----------------------------------------------------------------------------
105
106 #define BSWAP_UCS4(str, len) { unsigned _c; for (_c=0; _c<len; _c++) str[_c]=wxUINT32_SWAP_ALWAYS(str[_c]); }
107 #define BSWAP_UTF16(str, len) { unsigned _c; for (_c=0; _c<len; _c++) str[_c]=wxUINT16_SWAP_ALWAYS(str[_c]); }
108
109 // under Unix SIZEOF_WCHAR_T is defined by configure, but under other platforms
110 // it might be not defined - assume the most common value
111 #ifndef SIZEOF_WCHAR_T
112 #define SIZEOF_WCHAR_T 2
113 #endif // !defined(SIZEOF_WCHAR_T)
114
115 #if SIZEOF_WCHAR_T == 4
116 #define WC_NAME "UCS4"
117 #define WC_BSWAP BSWAP_UCS4
118 #ifdef WORDS_BIGENDIAN
119 #define WC_NAME_BEST "UCS-4BE"
120 #else
121 #define WC_NAME_BEST "UCS-4LE"
122 #endif
123 #elif SIZEOF_WCHAR_T == 2
124 #define WC_NAME "UTF16"
125 #define WC_BSWAP BSWAP_UTF16
126 #define WC_UTF16
127 #ifdef WORDS_BIGENDIAN
128 #define WC_NAME_BEST "UTF-16BE"
129 #else
130 #define WC_NAME_BEST "UTF-16LE"
131 #endif
132 #else // sizeof(wchar_t) != 2 nor 4
133 // I don't know what to do about this
134 #error "Weird sizeof(wchar_t): please report your platform details to wx-users mailing list"
135 #endif
136
137 // ============================================================================
138 // implementation
139 // ============================================================================
140
141 // ----------------------------------------------------------------------------
142 // UTF-16 en/decoding
143 // ----------------------------------------------------------------------------
144
145 #ifdef WC_UTF16
146
147 static size_t encode_utf16(wxUint32 input, wchar_t *output)
148 {
149 if (input<=0xffff)
150 {
151 if (output) *output++ = (wchar_t) input;
152 return 1;
153 }
154 else if (input>=0x110000)
155 {
156 return (size_t)-1;
157 }
158 else
159 {
160 if (output)
161 {
162 *output++ = (wchar_t) ((input >> 10)+0xd7c0);
163 *output++ = (wchar_t) ((input&0x3ff)+0xdc00);
164 }
165 return 2;
166 }
167 }
168
169 static size_t decode_utf16(const wchar_t* input, wxUint32& output)
170 {
171 if ((*input<0xd800) || (*input>0xdfff))
172 {
173 output = *input;
174 return 1;
175 }
176 else if ((input[1]<0xdc00) || (input[1]>=0xdfff))
177 {
178 output = *input;
179 return (size_t)-1;
180 }
181 else
182 {
183 output = ((input[0] - 0xd7c0) << 10) + (input[1] - 0xdc00);
184 return 2;
185 }
186 }
187
188 #endif // WC_UTF16
189
190 // ----------------------------------------------------------------------------
191 // wxMBConv
192 // ----------------------------------------------------------------------------
193
194 #define IGNORE_LIBC 0
195
196 wxMBConv::~wxMBConv()
197 {
198 // nothing to do here
199 }
200
201 size_t wxMBConv::MB2WC(wchar_t *buf, const char *psz, size_t n) const
202 {
203 #if IGNORE_LIBC
204 if (buf)
205 {
206 for (size_t i = 0; i < strlen( psz )+1; i++)
207 buf[i] = (wchar_t) psz[i];
208 return strlen( psz );
209 }
210 else
211 {
212 return strlen( psz );
213 }
214 #else
215 return wxMB2WC(buf, psz, n);
216 #endif
217 }
218
219 size_t wxMBConv::WC2MB(char *buf, const wchar_t *psz, size_t n) const
220 {
221 #if IGNORE_LIBC
222 if (buf)
223 {
224 for (size_t i = 0; i < wxStrlen( psz )+1; i++)
225 buf[i] = (char) psz[i];
226 return wxStrlen( psz );
227 }
228 else
229 {
230 return wxStrlen( psz );
231 }
232 #else
233 return wxWC2MB(buf, psz, n);
234 #endif
235 }
236
237 const wxWCharBuffer wxMBConv::cMB2WC(const char *psz) const
238 {
239 if ( psz )
240 {
241 // calculate the length of the buffer needed first
242 size_t nLen = MB2WC(NULL, psz, 0);
243 if ( nLen != (size_t)-1 )
244 {
245 // now do the actual conversion
246 wxWCharBuffer buf(nLen);
247 MB2WC(buf.data(), psz, nLen + 1); // with the trailing NUL
248
249 return buf;
250 }
251 }
252
253 wxWCharBuffer buf((wchar_t *)NULL);
254
255 return buf;
256 }
257
258 const wxCharBuffer wxMBConv::cWC2MB(const wchar_t *pwz) const
259 {
260 if ( pwz )
261 {
262 size_t nLen = WC2MB(NULL, pwz, 0);
263 if ( nLen != (size_t)-1 )
264 {
265 wxCharBuffer buf(nLen);
266 WC2MB(buf.data(), pwz, nLen + 1);
267
268 return buf;
269 }
270 }
271
272 wxCharBuffer buf((char *)NULL);
273
274 return buf;
275 }
276
277 // ----------------------------------------------------------------------------
278 // standard gdk conversion
279 // ----------------------------------------------------------------------------
280
281 #ifdef __WXGTK12__
282
283 WXDLLEXPORT_DATA(wxMBConvGdk) wxConvGdk;
284
285 #include <gdk/gdk.h>
286
287 size_t wxMBConvGdk::MB2WC(wchar_t *buf, const char *psz, size_t n) const
288 {
289 if (buf)
290 {
291 return gdk_mbstowcs((GdkWChar *)buf, psz, n);
292 }
293 else
294 {
295 GdkWChar *nbuf = new GdkWChar[n=strlen(psz)];
296 size_t len = gdk_mbstowcs(nbuf, psz, n);
297 delete[] nbuf;
298 return len;
299 }
300 }
301
302 size_t wxMBConvGdk::WC2MB(char *buf, const wchar_t *psz, size_t n) const
303 {
304 char *mbstr = gdk_wcstombs((GdkWChar *)psz);
305 size_t len = mbstr ? strlen(mbstr) : 0;
306 if (buf)
307 {
308 if (len > n)
309 len = n;
310 memcpy(buf, psz, len);
311 if (len < n)
312 buf[len] = 0;
313 }
314 return len;
315 }
316
317 #endif // GTK > 1.0
318
319 // ----------------------------------------------------------------------------
320 // UTF-7
321 // ----------------------------------------------------------------------------
322
323 WXDLLEXPORT_DATA(wxMBConvUTF7) wxConvUTF7;
324
325 #if 0
326 static char utf7_setD[]="ABCDEFGHIJKLMNOPQRSTUVWXYZ"
327 "abcdefghijklmnopqrstuvwxyz"
328 "0123456789'(),-./:?";
329 static char utf7_setO[]="!\"#$%&*;<=>@[]^_`{|}";
330 static char utf7_setB[]="ABCDEFGHIJKLMNOPQRSTUVWXYZ"
331 "abcdefghijklmnopqrstuvwxyz"
332 "0123456789+/";
333 #endif
334
335 // TODO: write actual implementations of UTF-7 here
336 size_t wxMBConvUTF7::MB2WC(wchar_t * WXUNUSED(buf),
337 const char * WXUNUSED(psz),
338 size_t WXUNUSED(n)) const
339 {
340 return 0;
341 }
342
343 size_t wxMBConvUTF7::WC2MB(char * WXUNUSED(buf),
344 const wchar_t * WXUNUSED(psz),
345 size_t WXUNUSED(n)) const
346 {
347 return 0;
348 }
349
350 // ----------------------------------------------------------------------------
351 // UTF-8
352 // ----------------------------------------------------------------------------
353
354 WXDLLEXPORT_DATA(wxMBConvUTF8) wxConvUTF8;
355
356 static wxUint32 utf8_max[]=
357 { 0x7f, 0x7ff, 0xffff, 0x1fffff, 0x3ffffff, 0x7fffffff, 0xffffffff };
358
359 size_t wxMBConvUTF8::MB2WC(wchar_t *buf, const char *psz, size_t n) const
360 {
361 size_t len = 0;
362
363 while (*psz && ((!buf) || (len < n)))
364 {
365 unsigned char cc = *psz++, fc = cc;
366 unsigned cnt;
367 for (cnt = 0; fc & 0x80; cnt++)
368 fc <<= 1;
369 if (!cnt)
370 {
371 // plain ASCII char
372 if (buf)
373 *buf++ = cc;
374 len++;
375 }
376 else
377 {
378 cnt--;
379 if (!cnt)
380 {
381 // invalid UTF-8 sequence
382 return (size_t)-1;
383 }
384 else
385 {
386 unsigned ocnt = cnt - 1;
387 wxUint32 res = cc & (0x3f >> cnt);
388 while (cnt--)
389 {
390 cc = *psz++;
391 if ((cc & 0xC0) != 0x80)
392 {
393 // invalid UTF-8 sequence
394 return (size_t)-1;
395 }
396 res = (res << 6) | (cc & 0x3f);
397 }
398 if (res <= utf8_max[ocnt])
399 {
400 // illegal UTF-8 encoding
401 return (size_t)-1;
402 }
403 #ifdef WC_UTF16
404 size_t pa = encode_utf16(res, buf);
405 if (pa == (size_t)-1)
406 return (size_t)-1;
407 if (buf)
408 buf += pa;
409 len += pa;
410 #else // !WC_UTF16
411 if (buf)
412 *buf++ = res;
413 len++;
414 #endif // WC_UTF16/!WC_UTF16
415 }
416 }
417 }
418 if (buf && (len < n))
419 *buf = 0;
420 return len;
421 }
422
423 size_t wxMBConvUTF8::WC2MB(char *buf, const wchar_t *psz, size_t n) const
424 {
425 size_t len = 0;
426
427 while (*psz && ((!buf) || (len < n)))
428 {
429 wxUint32 cc;
430 #ifdef WC_UTF16
431 size_t pa = decode_utf16(psz, cc);
432 psz += (pa == (size_t)-1) ? 1 : pa;
433 #else
434 cc=(*psz++) & 0x7fffffff;
435 #endif
436 unsigned cnt;
437 for (cnt = 0; cc > utf8_max[cnt]; cnt++) {}
438 if (!cnt)
439 {
440 // plain ASCII char
441 if (buf)
442 *buf++ = (char) cc;
443 len++;
444 }
445
446 else
447 {
448 len += cnt + 1;
449 if (buf)
450 {
451 *buf++ = (char) ((-128 >> cnt) | ((cc >> (cnt * 6)) & (0x3f >> cnt)));
452 while (cnt--)
453 *buf++ = (char) (0x80 | ((cc >> (cnt * 6)) & 0x3f));
454 }
455 }
456 }
457
458 if (buf && (len<n)) *buf = 0;
459
460 return len;
461 }
462
463 // ============================================================================
464 // wxCharacterSet and derived classes
465 // ============================================================================
466
467 // ----------------------------------------------------------------------------
468 // wxCharacterSet is the ABC for the classes below
469 // ----------------------------------------------------------------------------
470
471 class wxCharacterSet
472 {
473 public:
474 wxCharacterSet(const wxChar*name) : cname(name) {}
475 virtual ~wxCharacterSet() {}
476 virtual size_t MB2WC(wchar_t *buf, const char *psz, size_t n) = 0;
477 virtual size_t WC2MB(char *buf, const wchar_t *psz, size_t n) = 0;
478 virtual bool usable() const = 0;
479 public:
480 const wxChar*cname;
481 };
482
483 // ----------------------------------------------------------------------------
484 // ID_CharSet: implementation of wxCharacterSet using an existing wxMBConv
485 // ----------------------------------------------------------------------------
486
487 class ID_CharSet : public wxCharacterSet
488 {
489 public:
490 ID_CharSet(const wxChar *name, wxMBConv *cnv)
491 : wxCharacterSet(name), work(cnv) {}
492
493 size_t MB2WC(wchar_t *buf, const char *psz, size_t n)
494 { return work ? work->MB2WC(buf,psz,n) : (size_t)-1; }
495
496 size_t WC2MB(char *buf, const wchar_t *psz, size_t n)
497 { return work ? work->WC2MB(buf,psz,n) : (size_t)-1; }
498
499 bool usable() const
500 { return work!=NULL; }
501 public:
502 wxMBConv*work;
503 };
504
505
506 // ============================================================================
507 // The classes doing conversion using the iconv_xxx() functions
508 // ============================================================================
509
510 #ifdef HAVE_ICONV
511
512 // VS: glibc 2.1.3 is broken in that iconv() conversion to/from UCS4 fails with E2BIG
513 // if output buffer is _exactly_ as big as needed. Such case is (unless there's
514 // yet another bug in glibc) the only case when iconv() returns with (size_t)-1
515 // (which means error) and says there are 0 bytes left in the input buffer --
516 // when _real_ error occurs, bytes-left-in-input buffer is non-zero. Hence,
517 // this alternative test for iconv() failure.
518 // [This bug does not appear in glibc 2.2.]
519 #if defined(__GLIBC__) && __GLIBC__ == 2 && __GLIBC_MINOR__ <= 1
520 #define ICONV_FAILED(cres, bufLeft) ((cres == (size_t)-1) && \
521 (errno != E2BIG || bufLeft != 0))
522 #else
523 #define ICONV_FAILED(cres, bufLeft) (cres == (size_t)-1)
524 #endif
525
526 #define ICONV_CHAR_CAST(x) ((ICONV_CONST char **)(x))
527
528 // ----------------------------------------------------------------------------
529 // IC_CharSet: encapsulates an iconv character set
530 // ----------------------------------------------------------------------------
531
532 class IC_CharSet : public wxCharacterSet
533 {
534 public:
535 IC_CharSet(const wxChar *name);
536 virtual ~IC_CharSet();
537
538 virtual size_t MB2WC(wchar_t *buf, const char *psz, size_t n);
539 virtual size_t WC2MB(char *buf, const wchar_t *psz, size_t n);
540
541 bool usable() const
542 { return (m2w != (iconv_t)-1) && (w2m != (iconv_t)-1); }
543
544 protected:
545 // the iconv handlers used to translate from multibyte to wide char and in
546 // the other direction
547 iconv_t m2w,
548 w2m;
549
550 private:
551 // the name (for iconv_open()) of a wide char charset - if none is
552 // available on this machine, it will remain NULL
553 static const char *ms_wcCharsetName;
554
555 // true if the wide char encoding we use (i.e. ms_wcCharsetName) has
556 // different endian-ness than the native one
557 static bool ms_wcNeedsSwap;
558 };
559
560 const char *IC_CharSet::ms_wcCharsetName = NULL;
561 bool IC_CharSet::ms_wcNeedsSwap = FALSE;
562
563 IC_CharSet::IC_CharSet(const wxChar *name)
564 : wxCharacterSet(name)
565 {
566 // Do it the hard way
567 char cname[100];
568 for (size_t i = 0; i < wxStrlen(name)+1; i++)
569 cname[i] = (char) name[i];
570
571 // check for charset that represents wchar_t:
572 if (ms_wcCharsetName == NULL)
573 {
574 ms_wcNeedsSwap = FALSE;
575
576 // try charset with explicit bytesex info (e.g. "UCS-4LE"):
577 ms_wcCharsetName = WC_NAME_BEST;
578 m2w = iconv_open(ms_wcCharsetName, cname);
579
580 if (m2w == (iconv_t)-1)
581 {
582 // try charset w/o bytesex info (e.g. "UCS4")
583 // and check for bytesex ourselves:
584 ms_wcCharsetName = WC_NAME;
585 m2w = iconv_open(ms_wcCharsetName, cname);
586
587 // last bet, try if it knows WCHAR_T pseudo-charset
588 if (m2w == (iconv_t)-1)
589 {
590 ms_wcCharsetName = "WCHAR_T";
591 m2w = iconv_open(ms_wcCharsetName, cname);
592 }
593
594 if (m2w != (iconv_t)-1)
595 {
596 char buf[2], *bufPtr;
597 wchar_t wbuf[2], *wbufPtr;
598 size_t insz, outsz;
599 size_t res;
600
601 buf[0] = 'A';
602 buf[1] = 0;
603 wbuf[0] = 0;
604 insz = 2;
605 outsz = SIZEOF_WCHAR_T * 2;
606 wbufPtr = wbuf;
607 bufPtr = buf;
608
609 res = iconv(m2w, ICONV_CHAR_CAST(&bufPtr), &insz,
610 (char**)&wbufPtr, &outsz);
611
612 if (ICONV_FAILED(res, insz))
613 {
614 ms_wcCharsetName = NULL;
615 wxLogLastError(wxT("iconv"));
616 wxLogError(_("Conversion to charset '%s' doesn't work."), name);
617 }
618 else
619 {
620 ms_wcNeedsSwap = wbuf[0] != (wchar_t)buf[0];
621 }
622 }
623 else
624 {
625 ms_wcCharsetName = NULL;
626
627 // VS: we must not output an error here, since wxWindows will safely
628 // fall back to using wxEncodingConverter.
629 wxLogTrace(wxT("strconv"), wxT("Impossible to convert to/from charset '%s' with iconv, falling back to wxEncodingConverter."), name);
630 //wxLogError(
631 }
632 }
633 wxLogTrace(wxT("strconv"), wxT("wchar_t charset is '%s', needs swap: %i"), ms_wcCharsetName, ms_wcNeedsSwap);
634 }
635 else // we already have ms_wcCharsetName
636 {
637 m2w = iconv_open(ms_wcCharsetName, cname);
638 }
639
640 // NB: don't ever pass NULL to iconv_open(), it may crash!
641 if ( ms_wcCharsetName )
642 {
643 w2m = iconv_open( cname, ms_wcCharsetName);
644 }
645 else
646 {
647 w2m = (iconv_t)-1;
648 }
649 }
650
651 IC_CharSet::~IC_CharSet()
652 {
653 if ( m2w != (iconv_t)-1 )
654 iconv_close(m2w);
655 if ( w2m != (iconv_t)-1 )
656 iconv_close(w2m);
657 }
658
659 size_t IC_CharSet::MB2WC(wchar_t *buf, const char *psz, size_t n)
660 {
661 size_t inbuf = strlen(psz);
662 size_t outbuf = n * SIZEOF_WCHAR_T;
663 size_t res, cres;
664 // VS: Use these instead of psz, buf because iconv() modifies its arguments:
665 wchar_t *bufPtr = buf;
666 const char *pszPtr = psz;
667
668 if (buf)
669 {
670 // have destination buffer, convert there
671 cres = iconv(m2w,
672 ICONV_CHAR_CAST(&pszPtr), &inbuf,
673 (char**)&bufPtr, &outbuf);
674 res = n - (outbuf / SIZEOF_WCHAR_T);
675
676 if (ms_wcNeedsSwap)
677 {
678 // convert to native endianness
679 WC_BSWAP(buf /* _not_ bufPtr */, res)
680 }
681
682 // NB: iconv was given only strlen(psz) characters on input, and so
683 // it couldn't convert the trailing zero. Let's do it ourselves
684 // if there's some room left for it in the output buffer.
685 if (res < n)
686 buf[res] = 0;
687 }
688 else
689 {
690 // no destination buffer... convert using temp buffer
691 // to calculate destination buffer requirement
692 wchar_t tbuf[8];
693 res = 0;
694 do {
695 bufPtr = tbuf;
696 outbuf = 8*SIZEOF_WCHAR_T;
697
698 cres = iconv(m2w,
699 ICONV_CHAR_CAST(&pszPtr), &inbuf,
700 (char**)&bufPtr, &outbuf );
701
702 res += 8-(outbuf/SIZEOF_WCHAR_T);
703 } while ((cres==(size_t)-1) && (errno==E2BIG));
704 }
705
706 if (ICONV_FAILED(cres, inbuf))
707 {
708 //VS: it is ok if iconv fails, hence trace only
709 wxLogTrace(wxT("strconv"), wxT("iconv failed: %s"), wxSysErrorMsg(wxSysErrorCode()));
710 return (size_t)-1;
711 }
712
713 return res;
714 }
715
716 size_t IC_CharSet::WC2MB(char *buf, const wchar_t *psz, size_t n)
717 {
718 size_t inbuf = wxWcslen(psz) * SIZEOF_WCHAR_T;
719 size_t outbuf = n;
720 size_t res, cres;
721
722 wchar_t *tmpbuf = 0;
723
724 if (ms_wcNeedsSwap)
725 {
726 // need to copy to temp buffer to switch endianness
727 // this absolutely doesn't rock!
728 // (no, doing WC_BSWAP twice on the original buffer won't help, as it
729 // could be in read-only memory, or be accessed in some other thread)
730 tmpbuf=(wchar_t*)malloc((inbuf+1)*SIZEOF_WCHAR_T);
731 memcpy(tmpbuf,psz,(inbuf+1)*SIZEOF_WCHAR_T);
732 WC_BSWAP(tmpbuf, inbuf)
733 psz=tmpbuf;
734 }
735
736 if (buf)
737 {
738 // have destination buffer, convert there
739 cres = iconv( w2m, ICONV_CHAR_CAST(&psz), &inbuf, &buf, &outbuf );
740
741 res = n-outbuf;
742
743 // NB: iconv was given only wcslen(psz) characters on input, and so
744 // it couldn't convert the trailing zero. Let's do it ourselves
745 // if there's some room left for it in the output buffer.
746 if (res < n)
747 buf[0] = 0;
748 }
749 else
750 {
751 // no destination buffer... convert using temp buffer
752 // to calculate destination buffer requirement
753 char tbuf[16];
754 res = 0;
755 do {
756 buf = tbuf; outbuf = 16;
757
758 cres = iconv( w2m, ICONV_CHAR_CAST(&psz), &inbuf, &buf, &outbuf );
759
760 res += 16 - outbuf;
761 } while ((cres==(size_t)-1) && (errno==E2BIG));
762 }
763
764 if (ms_wcNeedsSwap)
765 {
766 free(tmpbuf);
767 }
768
769 if (ICONV_FAILED(cres, inbuf))
770 {
771 //VS: it is ok if iconv fails, hence trace only
772 wxLogTrace(wxT("strconv"), wxT("iconv failed: %s"), wxSysErrorMsg(wxSysErrorCode()));
773 return (size_t)-1;
774 }
775
776 return res;
777 }
778
779 #endif // HAVE_ICONV
780
781 // ============================================================================
782 // Win32 conversion classes
783 // ============================================================================
784
785 #if defined(__WIN32__) && !defined(__WXMICROWIN__) && !defined(__WXUNIVERSAL__)
786
787 extern long wxCharsetToCodepage(const wxChar *charset); // from utils.cpp
788
789 class CP_CharSet : public wxCharacterSet
790 {
791 public:
792 CP_CharSet(const wxChar* name)
793 : wxCharacterSet(name)
794 {
795 m_CodePage = wxCharsetToCodepage(name);
796 }
797
798 size_t MB2WC(wchar_t *buf, const char *psz, size_t n)
799 {
800 const size_t len = ::MultiByteToWideChar
801 (
802 m_CodePage, // code page
803 0, // flags (none)
804 psz, // input string
805 -1, // its length (NUL-terminated)
806 buf, // output string
807 buf ? n : 0 // size of output buffer
808 );
809
810 // note that it returns # of written chars for buf != NULL and *size*
811 // of the needed buffer for buf == NULL
812 return len ? (buf ? len : len - 1) : (size_t)-1;
813 }
814
815 size_t WC2MB(char *buf, const wchar_t *psz, size_t n)
816 {
817 const size_t len = ::WideCharToMultiByte
818 (
819 m_CodePage, // code page
820 0, // flags (none)
821 psz, // input string
822 -1, // it is (wide) NUL-terminated
823 buf, // output buffer
824 buf ? n : 0, // and its size
825 NULL, // default "replacement" char
826 NULL // [out] was it used?
827 );
828
829 // see the comment above!
830 return len ? (buf ? len : len - 1) : (size_t)-1;
831 }
832
833 bool usable() const
834 { return m_CodePage != -1; }
835
836 public:
837 long m_CodePage;
838 };
839 #endif // defined(__WIN32__) && !defined(__WXMICROWIN__) && !defined(__WXUNIVERSAL__)
840
841 // ============================================================================
842 // wxEncodingConverter based conversion classes
843 // ============================================================================
844
845 #if wxUSE_FONTMAP
846
847 class EC_CharSet : public wxCharacterSet
848 {
849 public:
850 // temporarily just use wxEncodingConverter stuff,
851 // so that it works while a better implementation is built
852 EC_CharSet(const wxChar* name) : wxCharacterSet(name),
853 enc(wxFONTENCODING_SYSTEM)
854 {
855 if (name)
856 enc = wxFontMapper::Get()->CharsetToEncoding(name, FALSE);
857
858 m_ok = m2w.Init(enc, wxFONTENCODING_UNICODE) &&
859 w2m.Init(wxFONTENCODING_UNICODE, enc);
860 }
861
862 size_t MB2WC(wchar_t *buf, const char *psz, size_t WXUNUSED(n))
863 {
864 size_t inbuf = strlen(psz);
865 if (buf)
866 m2w.Convert(psz,buf);
867 return inbuf;
868 }
869
870 size_t WC2MB(char *buf, const wchar_t *psz, size_t WXUNUSED(n))
871 {
872 const size_t inbuf = wxWcslen(psz);
873 if (buf)
874 w2m.Convert(psz,buf);
875
876 return inbuf;
877 }
878
879 bool usable() const { return m_ok; }
880
881 public:
882 wxFontEncoding enc;
883 wxEncodingConverter m2w, w2m;
884
885 // were we initialized successfully?
886 bool m_ok;
887 };
888
889 #endif // wxUSE_FONTMAP
890
891 // ----------------------------------------------------------------------------
892 // the function creating the wxCharacterSet for the specified charset on the
893 // current system, trying all possibilities
894 // ----------------------------------------------------------------------------
895
896 static wxCharacterSet *wxGetCharacterSet(const wxChar *name)
897 {
898 // check for the special case of ASCII charset
899 #if wxUSE_FONTMAP
900 if ( wxFontMapper::Get()->CharsetToEncoding(name) == wxFONTENCODING_DEFAULT )
901 #else // wxUSE_FONTMAP
902 if ( !name )
903 #endif // wxUSE_FONTMAP/!wxUSE_FONTMAP
904 {
905 // don't convert at all
906 return NULL;
907 }
908
909 // the test above must have taken care of this case
910 wxCHECK_MSG( name, NULL, _T("NULL name must be wxFONTENCODING_DEFAULT") );
911
912 wxCharacterSet *cset;
913
914 if ( wxStricmp(name, wxT("UTF8")) == 0 || wxStricmp(name, wxT("UTF-8")) == 0)
915 {
916 cset = new ID_CharSet(name, &wxConvUTF8);
917 }
918 else
919 {
920 #ifdef HAVE_ICONV
921 cset = new IC_CharSet(name);
922 #else // !HAVE_ICONV
923 cset = NULL;
924 #endif // HAVE_ICONV/!HAVE_ICONV
925 }
926
927 // it can only be NULL in this case
928 #ifndef HAVE_ICONV
929 if ( cset )
930 #endif // !HAVE_ICONV
931 {
932 if ( cset->usable() )
933 return cset;
934
935 delete cset;
936 cset = NULL;
937 }
938
939 #if defined(__WIN32__) && !defined(__WXMICROWIN__) && !defined(__WXUNIVERSAL__)
940 cset = new CP_CharSet(name);
941 if ( cset->usable() )
942 return cset;
943
944 delete cset;
945 cset = NULL;
946 #endif // defined(__WIN32__) && !defined(__WXMICROWIN__) && !defined(__WXUNIVERSAL__)
947
948 #if wxUSE_FONTMAP
949 cset = new EC_CharSet(name);
950 if ( cset->usable() )
951 return cset;
952
953 delete cset;
954 cset = NULL;
955 #endif // wxUSE_FONTMAP
956
957 wxLogError(_("Cannot convert from encoding '%s'!"), name);
958
959 return NULL;
960 }
961
962 // ============================================================================
963 // wxCSConv implementation
964 // ============================================================================
965
966 wxCSConv::wxCSConv(const wxChar *charset)
967 {
968 m_name = (wxChar *)NULL;
969 m_cset = (wxCharacterSet *) NULL;
970 m_deferred = TRUE;
971
972 SetName(charset);
973 }
974
975 wxCSConv::~wxCSConv()
976 {
977 Clear();
978 }
979
980 wxCSConv::wxCSConv(const wxCSConv& conv)
981 : wxMBConv()
982 {
983 Clear();
984 SetName(conv.m_name);
985 }
986
987 wxCSConv& wxCSConv::operator=(const wxCSConv& conv)
988 {
989 Clear();
990 SetName(conv.m_name);
991 return *this;
992 }
993
994 void wxCSConv::Clear()
995 {
996 if (m_name)
997 free(m_name);
998 if (m_cset)
999 delete m_cset;
1000 m_name = NULL;
1001 m_cset = NULL;
1002 }
1003
1004 void wxCSConv::SetName(const wxChar *charset)
1005 {
1006 if (charset)
1007 {
1008 m_name = wxStrdup(charset);
1009 m_deferred = TRUE;
1010 }
1011 }
1012
1013 void wxCSConv::LoadNow()
1014 {
1015 if ( m_deferred )
1016 {
1017 // it would probably be better to make GetSystemEncodingName() always
1018 // available (i.e. even when wxUSE_INTL == 0)?
1019 #if wxUSE_INTL
1020 if ( !m_name )
1021 {
1022 wxString name = wxLocale::GetSystemEncodingName();
1023 if ( !name.empty() )
1024 {
1025 SetName(name);
1026 }
1027 }
1028 #endif // wxUSE_INTL
1029
1030 // wxGetCharacterSet() complains about NULL name
1031 m_cset = m_name ? wxGetCharacterSet(m_name) : NULL;
1032 m_deferred = FALSE;
1033 }
1034 }
1035
1036 size_t wxCSConv::MB2WC(wchar_t *buf, const char *psz, size_t n) const
1037 {
1038 ((wxCSConv *)this)->LoadNow(); // discard constness
1039
1040 if (m_cset)
1041 return m_cset->MB2WC(buf, psz, n);
1042
1043 // latin-1 (direct)
1044 size_t len = strlen(psz);
1045
1046 if (buf)
1047 {
1048 for (size_t c = 0; c <= len; c++)
1049 buf[c] = (unsigned char)(psz[c]);
1050 }
1051
1052 return len;
1053 }
1054
1055 size_t wxCSConv::WC2MB(char *buf, const wchar_t *psz, size_t n) const
1056 {
1057 ((wxCSConv *)this)->LoadNow(); // discard constness
1058
1059 if (m_cset)
1060 return m_cset->WC2MB(buf, psz, n);
1061
1062 // latin-1 (direct)
1063 const size_t len = wxWcslen(psz);
1064 if (buf)
1065 {
1066 for (size_t c = 0; c <= len; c++)
1067 buf[c] = (psz[c] > 0xff) ? '?' : psz[c];
1068 }
1069
1070 return len;
1071 }
1072
1073 #endif // wxUSE_WCHAR_T
1074
1075
wxWidgets (Impactor)