projects / wxWidgets.git / blob
? search:
summary | shortlog | log | commit | commitdiff | tree
blame | history | raw | HEAD
Fixed copyrights and licence spelling
[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 licence
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 wxConvUTF8;
63 #endif // wxUSE_WCHAR_T
64
65 WXDLLEXPORT_DATA(wxMBConv *) wxConvCurrent = &wxConvLibc;
66
67 class wxStrConvModule: public wxModule
68 {
69 public:
70 wxStrConvModule() : wxModule() { }
71 virtual bool OnInit() { return TRUE; }
72 virtual void OnExit()
73 {
74 #if wxUSE_WCHAR_T
75 wxConvLocal.Clear();
76 wxConvISO8859_1.Clear();
77 #endif
78 }
79
80 DECLARE_DYNAMIC_CLASS(wxStrConvModule)
81 };
82
83 IMPLEMENT_DYNAMIC_CLASS(wxStrConvModule, wxModule)
84
85
86 // ----------------------------------------------------------------------------
87 // headers
88 // ----------------------------------------------------------------------------
89
90 #if wxUSE_WCHAR_T
91
92 #ifdef __SALFORDC__
93 #include <clib.h>
94 #endif
95
96 #ifdef HAVE_ICONV
97 #include <iconv.h>
98 #endif
99
100 #include "wx/encconv.h"
101 #include "wx/fontmap.h"
102
103 // ----------------------------------------------------------------------------
104 // macros
105 // ----------------------------------------------------------------------------
106
107 #define BSWAP_UCS4(str, len) { unsigned _c; for (_c=0; _c<len; _c++) str[_c]=wxUINT32_SWAP_ALWAYS(str[_c]); }
108 #define BSWAP_UTF16(str, len) { unsigned _c; for (_c=0; _c<len; _c++) str[_c]=wxUINT16_SWAP_ALWAYS(str[_c]); }
109
110 // under Unix SIZEOF_WCHAR_T is defined by configure, but under other platforms
111 // it might be not defined - assume the most common value
112 #ifndef SIZEOF_WCHAR_T
113 #define SIZEOF_WCHAR_T 2
114 #endif // !defined(SIZEOF_WCHAR_T)
115
116 #if SIZEOF_WCHAR_T == 4
117 #define WC_NAME "UCS4"
118 #define WC_BSWAP BSWAP_UCS4
119 #ifdef WORDS_BIGENDIAN
120 #define WC_NAME_BEST "UCS-4BE"
121 #else
122 #define WC_NAME_BEST "UCS-4LE"
123 #endif
124 #elif SIZEOF_WCHAR_T == 2
125 #define WC_NAME "UTF16"
126 #define WC_BSWAP BSWAP_UTF16
127 #define WC_UTF16
128 #ifdef WORDS_BIGENDIAN
129 #define WC_NAME_BEST "UTF-16BE"
130 #else
131 #define WC_NAME_BEST "UTF-16LE"
132 #endif
133 #else // sizeof(wchar_t) != 2 nor 4
134 // I don't know what to do about this
135 #error "Weird sizeof(wchar_t): please report your platform details to wx-users mailing list"
136 #endif
137
138 // ============================================================================
139 // implementation
140 // ============================================================================
141
142 // ----------------------------------------------------------------------------
143 // UTF-16 en/decoding
144 // ----------------------------------------------------------------------------
145
146 #ifdef WC_UTF16
147
148 static size_t encode_utf16(wxUint32 input, wchar_t *output)
149 {
150 if (input<=0xffff)
151 {
152 if (output) *output++ = (wchar_t) input;
153 return 1;
154 }
155 else if (input>=0x110000)
156 {
157 return (size_t)-1;
158 }
159 else
160 {
161 if (output)
162 {
163 *output++ = (wchar_t) ((input >> 10)+0xd7c0);
164 *output++ = (wchar_t) ((input&0x3ff)+0xdc00);
165 }
166 return 2;
167 }
168 }
169
170 static size_t decode_utf16(const wchar_t* input, wxUint32& output)
171 {
172 if ((*input<0xd800) || (*input>0xdfff))
173 {
174 output = *input;
175 return 1;
176 }
177 else if ((input[1]<0xdc00) || (input[1]>=0xdfff))
178 {
179 output = *input;
180 return (size_t)-1;
181 }
182 else
183 {
184 output = ((input[0] - 0xd7c0) << 10) + (input[1] - 0xdc00);
185 return 2;
186 }
187 }
188
189 #endif // WC_UTF16
190
191 // ----------------------------------------------------------------------------
192 // wxMBConv
193 // ----------------------------------------------------------------------------
194
195 #define IGNORE_LIBC 0
196
197 wxMBConv::~wxMBConv()
198 {
199 // nothing to do here
200 }
201
202 size_t wxMBConv::MB2WC(wchar_t *buf, const char *psz, size_t n) const
203 {
204 #if IGNORE_LIBC
205 if (buf)
206 {
207 for (size_t i = 0; i < strlen( psz )+1; i++)
208 buf[i] = (wchar_t) psz[i];
209 return strlen( psz );
210 }
211 else
212 {
213 return strlen( psz );
214 }
215 #else
216 return wxMB2WC(buf, psz, n);
217 #endif
218 }
219
220 size_t wxMBConv::WC2MB(char *buf, const wchar_t *psz, size_t n) const
221 {
222 #if IGNORE_LIBC
223 if (buf)
224 {
225 for (size_t i = 0; i < wxStrlen( psz )+1; i++)
226 buf[i] = (char) psz[i];
227 return wxStrlen( psz );
228 }
229 else
230 {
231 return wxStrlen( psz );
232 }
233 #else
234 return wxWC2MB(buf, psz, n);
235 #endif
236 }
237
238 const wxWCharBuffer wxMBConv::cMB2WC(const char *psz) const
239 {
240 if ( psz )
241 {
242 // calculate the length of the buffer needed first
243 size_t nLen = MB2WC(NULL, psz, 0);
244 if ( nLen != (size_t)-1 )
245 {
246 // now do the actual conversion
247 wxWCharBuffer buf(nLen);
248 MB2WC(buf.data(), psz, nLen + 1); // with the trailing NUL
249
250 return buf;
251 }
252 }
253
254 wxWCharBuffer buf((wchar_t *)NULL);
255
256 return buf;
257 }
258
259 const wxCharBuffer wxMBConv::cWC2MB(const wchar_t *pwz) const
260 {
261 if ( pwz )
262 {
263 size_t nLen = WC2MB(NULL, pwz, 0);
264 if ( nLen != (size_t)-1 )
265 {
266 wxCharBuffer buf(nLen);
267 WC2MB(buf.data(), pwz, nLen + 1);
268
269 return buf;
270 }
271 }
272
273 wxCharBuffer buf((char *)NULL);
274
275 return buf;
276 }
277
278 // ----------------------------------------------------------------------------
279 // standard gdk conversion
280 // ----------------------------------------------------------------------------
281
282 #ifdef __WXGTK12__
283
284 WXDLLEXPORT_DATA(wxMBConvGdk) wxConvGdk;
285
286 #include <gdk/gdk.h>
287
288 size_t wxMBConvGdk::MB2WC(wchar_t *buf, const char *psz, size_t n) const
289 {
290 if (buf)
291 {
292 return gdk_mbstowcs((GdkWChar *)buf, psz, n);
293 }
294 else
295 {
296 GdkWChar *nbuf = new GdkWChar[n=strlen(psz)];
297 size_t len = gdk_mbstowcs(nbuf, psz, n);
298 delete[] nbuf;
299 return len;
300 }
301 }
302
303 size_t wxMBConvGdk::WC2MB(char *buf, const wchar_t *psz, size_t n) const
304 {
305 char *mbstr = gdk_wcstombs((GdkWChar *)psz);
306 size_t len = mbstr ? strlen(mbstr) : 0;
307 if (buf)
308 {
309 if (len > n)
310 len = n;
311 memcpy(buf, psz, len);
312 if (len < n)
313 buf[len] = 0;
314 }
315 return len;
316 }
317
318 #endif // GTK > 1.0
319
320 // ----------------------------------------------------------------------------
321 // UTF-7
322 // ----------------------------------------------------------------------------
323
324 WXDLLEXPORT_DATA(wxMBConvUTF7) wxConvUTF7;
325
326 #if 0
327 static char utf7_setD[]="ABCDEFGHIJKLMNOPQRSTUVWXYZ"
328 "abcdefghijklmnopqrstuvwxyz"
329 "0123456789'(),-./:?";
330 static char utf7_setO[]="!\"#$%&*;<=>@[]^_`{|}";
331 static char utf7_setB[]="ABCDEFGHIJKLMNOPQRSTUVWXYZ"
332 "abcdefghijklmnopqrstuvwxyz"
333 "0123456789+/";
334 #endif
335
336 // TODO: write actual implementations of UTF-7 here
337 size_t wxMBConvUTF7::MB2WC(wchar_t * WXUNUSED(buf),
338 const char * WXUNUSED(psz),
339 size_t WXUNUSED(n)) const
340 {
341 return 0;
342 }
343
344 size_t wxMBConvUTF7::WC2MB(char * WXUNUSED(buf),
345 const wchar_t * WXUNUSED(psz),
346 size_t WXUNUSED(n)) const
347 {
348 return 0;
349 }
350
351 // ----------------------------------------------------------------------------
352 // UTF-8
353 // ----------------------------------------------------------------------------
354
355 WXDLLEXPORT_DATA(wxMBConvUTF8) wxConvUTF8;
356
357 static wxUint32 utf8_max[]=
358 { 0x7f, 0x7ff, 0xffff, 0x1fffff, 0x3ffffff, 0x7fffffff, 0xffffffff };
359
360 size_t wxMBConvUTF8::MB2WC(wchar_t *buf, const char *psz, size_t n) const
361 {
362 size_t len = 0;
363
364 while (*psz && ((!buf) || (len < n)))
365 {
366 unsigned char cc = *psz++, fc = cc;
367 unsigned cnt;
368 for (cnt = 0; fc & 0x80; cnt++)
369 fc <<= 1;
370 if (!cnt)
371 {
372 // plain ASCII char
373 if (buf)
374 *buf++ = cc;
375 len++;
376 }
377 else
378 {
379 cnt--;
380 if (!cnt)
381 {
382 // invalid UTF-8 sequence
383 return (size_t)-1;
384 }
385 else
386 {
387 unsigned ocnt = cnt - 1;
388 wxUint32 res = cc & (0x3f >> cnt);
389 while (cnt--)
390 {
391 cc = *psz++;
392 if ((cc & 0xC0) != 0x80)
393 {
394 // invalid UTF-8 sequence
395 return (size_t)-1;
396 }
397 res = (res << 6) | (cc & 0x3f);
398 }
399 if (res <= utf8_max[ocnt])
400 {
401 // illegal UTF-8 encoding
402 return (size_t)-1;
403 }
404 #ifdef WC_UTF16
405 size_t pa = encode_utf16(res, buf);
406 if (pa == (size_t)-1)
407 return (size_t)-1;
408 if (buf)
409 buf += pa;
410 len += pa;
411 #else // !WC_UTF16
412 if (buf)
413 *buf++ = res;
414 len++;
415 #endif // WC_UTF16/!WC_UTF16
416 }
417 }
418 }
419 if (buf && (len < n))
420 *buf = 0;
421 return len;
422 }
423
424 size_t wxMBConvUTF8::WC2MB(char *buf, const wchar_t *psz, size_t n) const
425 {
426 size_t len = 0;
427
428 while (*psz && ((!buf) || (len < n)))
429 {
430 wxUint32 cc;
431 #ifdef WC_UTF16
432 size_t pa = decode_utf16(psz, cc);
433 psz += (pa == (size_t)-1) ? 1 : pa;
434 #else
435 cc=(*psz++) & 0x7fffffff;
436 #endif
437 unsigned cnt;
438 for (cnt = 0; cc > utf8_max[cnt]; cnt++) {}
439 if (!cnt)
440 {
441 // plain ASCII char
442 if (buf)
443 *buf++ = (char) cc;
444 len++;
445 }
446
447 else
448 {
449 len += cnt + 1;
450 if (buf)
451 {
452 *buf++ = (char) ((-128 >> cnt) | ((cc >> (cnt * 6)) & (0x3f >> cnt)));
453 while (cnt--)
454 *buf++ = (char) (0x80 | ((cc >> (cnt * 6)) & 0x3f));
455 }
456 }
457 }
458
459 if (buf && (len<n)) *buf = 0;
460
461 return len;
462 }
463
464 // ============================================================================
465 // wxCharacterSet and derived classes
466 // ============================================================================
467
468 // ----------------------------------------------------------------------------
469 // wxCharacterSet is the ABC for the classes below
470 // ----------------------------------------------------------------------------
471
472 class wxCharacterSet
473 {
474 public:
475 wxCharacterSet(const wxChar*name) : cname(name) {}
476 virtual ~wxCharacterSet() {}
477 virtual size_t MB2WC(wchar_t *buf, const char *psz, size_t n) = 0;
478 virtual size_t WC2MB(char *buf, const wchar_t *psz, size_t n) = 0;
479 virtual bool usable() const = 0;
480 public:
481 const wxChar*cname;
482 };
483
484 // ----------------------------------------------------------------------------
485 // ID_CharSet: implementation of wxCharacterSet using an existing wxMBConv
486 // ----------------------------------------------------------------------------
487
488 class ID_CharSet : public wxCharacterSet
489 {
490 public:
491 ID_CharSet(const wxChar *name, wxMBConv *cnv)
492 : wxCharacterSet(name), work(cnv) {}
493
494 size_t MB2WC(wchar_t *buf, const char *psz, size_t n)
495 { return work ? work->MB2WC(buf,psz,n) : (size_t)-1; }
496
497 size_t WC2MB(char *buf, const wchar_t *psz, size_t n)
498 { return work ? work->WC2MB(buf,psz,n) : (size_t)-1; }
499
500 bool usable() const
501 { return work!=NULL; }
502 public:
503 wxMBConv*work;
504 };
505
506
507 // ============================================================================
508 // The classes doing conversion using the iconv_xxx() functions
509 // ============================================================================
510
511 #ifdef HAVE_ICONV
512
513 // VS: glibc 2.1.3 is broken in that iconv() conversion to/from UCS4 fails with E2BIG
514 // if output buffer is _exactly_ as big as needed. Such case is (unless there's
515 // yet another bug in glibc) the only case when iconv() returns with (size_t)-1
516 // (which means error) and says there are 0 bytes left in the input buffer --
517 // when _real_ error occurs, bytes-left-in-input buffer is non-zero. Hence,
518 // this alternative test for iconv() failure.
519 // [This bug does not appear in glibc 2.2.]
520 #if defined(__GLIBC__) && __GLIBC__ == 2 && __GLIBC_MINOR__ <= 1
521 #define ICONV_FAILED(cres, bufLeft) ((cres == (size_t)-1) && \
522 (errno != E2BIG || bufLeft != 0))
523 #else
524 #define ICONV_FAILED(cres, bufLeft) (cres == (size_t)-1)
525 #endif
526
527 #define ICONV_CHAR_CAST(x) ((ICONV_CONST char **)(x))
528
529 // ----------------------------------------------------------------------------
530 // IC_CharSet: encapsulates an iconv character set
531 // ----------------------------------------------------------------------------
532
533 class IC_CharSet : public wxCharacterSet
534 {
535 public:
536 IC_CharSet(const wxChar *name);
537 virtual ~IC_CharSet();
538
539 virtual size_t MB2WC(wchar_t *buf, const char *psz, size_t n);
540 virtual size_t WC2MB(char *buf, const wchar_t *psz, size_t n);
541
542 bool usable() const
543 { return (m2w != (iconv_t)-1) && (w2m != (iconv_t)-1); }
544
545 protected:
546 // the iconv handlers used to translate from multibyte to wide char and in
547 // the other direction
548 iconv_t m2w,
549 w2m;
550
551 private:
552 // the name (for iconv_open()) of a wide char charset - if none is
553 // available on this machine, it will remain NULL
554 static const char *ms_wcCharsetName;
555
556 // true if the wide char encoding we use (i.e. ms_wcCharsetName) has
557 // different endian-ness than the native one
558 static bool ms_wcNeedsSwap;
559 };
560
561 const char *IC_CharSet::ms_wcCharsetName = NULL;
562 bool IC_CharSet::ms_wcNeedsSwap = FALSE;
563
564 IC_CharSet::IC_CharSet(const wxChar *name)
565 : wxCharacterSet(name)
566 {
567 // Do it the hard way
568 char cname[100];
569 for (size_t i = 0; i < wxStrlen(name)+1; i++)
570 cname[i] = (char) name[i];
571
572 // check for charset that represents wchar_t:
573 if (ms_wcCharsetName == NULL)
574 {
575 ms_wcNeedsSwap = FALSE;
576
577 // try charset with explicit bytesex info (e.g. "UCS-4LE"):
578 ms_wcCharsetName = WC_NAME_BEST;
579 m2w = iconv_open(ms_wcCharsetName, cname);
580
581 if (m2w == (iconv_t)-1)
582 {
583 // try charset w/o bytesex info (e.g. "UCS4")
584 // and check for bytesex ourselves:
585 ms_wcCharsetName = WC_NAME;
586 m2w = iconv_open(ms_wcCharsetName, cname);
587
588 // last bet, try if it knows WCHAR_T pseudo-charset
589 if (m2w == (iconv_t)-1)
590 {
591 ms_wcCharsetName = "WCHAR_T";
592 m2w = iconv_open(ms_wcCharsetName, cname);
593 }
594
595 if (m2w != (iconv_t)-1)
596 {
597 char buf[2], *bufPtr;
598 wchar_t wbuf[2], *wbufPtr;
599 size_t insz, outsz;
600 size_t res;
601
602 buf[0] = 'A';
603 buf[1] = 0;
604 wbuf[0] = 0;
605 insz = 2;
606 outsz = SIZEOF_WCHAR_T * 2;
607 wbufPtr = wbuf;
608 bufPtr = buf;
609
610 res = iconv(m2w, ICONV_CHAR_CAST(&bufPtr), &insz,
611 (char**)&wbufPtr, &outsz);
612
613 if (ICONV_FAILED(res, insz))
614 {
615 ms_wcCharsetName = NULL;
616 wxLogLastError(wxT("iconv"));
617 wxLogError(_("Conversion to charset '%s' doesn't work."), name);
618 }
619 else
620 {
621 ms_wcNeedsSwap = wbuf[0] != (wchar_t)buf[0];
622 }
623 }
624 else
625 {
626 ms_wcCharsetName = NULL;
627
628 // VS: we must not output an error here, since wxWindows will safely
629 // fall back to using wxEncodingConverter.
630 wxLogTrace(wxT("strconv"), wxT("Impossible to convert to/from charset '%s' with iconv, falling back to wxEncodingConverter."), name);
631 //wxLogError(
632 }
633 }
634 wxLogTrace(wxT("strconv"), wxT("wchar_t charset is '%s', needs swap: %i"), ms_wcCharsetName, ms_wcNeedsSwap);
635 }
636 else // we already have ms_wcCharsetName
637 {
638 m2w = iconv_open(ms_wcCharsetName, cname);
639 }
640
641 // NB: don't ever pass NULL to iconv_open(), it may crash!
642 if ( ms_wcCharsetName )
643 {
644 w2m = iconv_open( cname, ms_wcCharsetName);
645 }
646 else
647 {
648 w2m = (iconv_t)-1;
649 }
650 }
651
652 IC_CharSet::~IC_CharSet()
653 {
654 if ( m2w != (iconv_t)-1 )
655 iconv_close(m2w);
656 if ( w2m != (iconv_t)-1 )
657 iconv_close(w2m);
658 }
659
660 size_t IC_CharSet::MB2WC(wchar_t *buf, const char *psz, size_t n)
661 {
662 size_t inbuf = strlen(psz);
663 size_t outbuf = n * SIZEOF_WCHAR_T;
664 size_t res, cres;
665 // VS: Use these instead of psz, buf because iconv() modifies its arguments:
666 wchar_t *bufPtr = buf;
667 const char *pszPtr = psz;
668
669 if (buf)
670 {
671 // have destination buffer, convert there
672 cres = iconv(m2w,
673 ICONV_CHAR_CAST(&pszPtr), &inbuf,
674 (char**)&bufPtr, &outbuf);
675 res = n - (outbuf / SIZEOF_WCHAR_T);
676
677 if (ms_wcNeedsSwap)
678 {
679 // convert to native endianness
680 WC_BSWAP(buf /* _not_ bufPtr */, res)
681 }
682
683 // NB: iconv was given only strlen(psz) characters on input, and so
684 // it couldn't convert the trailing zero. Let's do it ourselves
685 // if there's some room left for it in the output buffer.
686 if (res < n)
687 buf[res] = 0;
688 }
689 else
690 {
691 // no destination buffer... convert using temp buffer
692 // to calculate destination buffer requirement
693 wchar_t tbuf[8];
694 res = 0;
695 do {
696 bufPtr = tbuf;
697 outbuf = 8*SIZEOF_WCHAR_T;
698
699 cres = iconv(m2w,
700 ICONV_CHAR_CAST(&pszPtr), &inbuf,
701 (char**)&bufPtr, &outbuf );
702
703 res += 8-(outbuf/SIZEOF_WCHAR_T);
704 } while ((cres==(size_t)-1) && (errno==E2BIG));
705 }
706
707 if (ICONV_FAILED(cres, inbuf))
708 {
709 //VS: it is ok if iconv fails, hence trace only
710 wxLogTrace(wxT("strconv"), wxT("iconv failed: %s"), wxSysErrorMsg(wxSysErrorCode()));
711 return (size_t)-1;
712 }
713
714 return res;
715 }
716
717 size_t IC_CharSet::WC2MB(char *buf, const wchar_t *psz, size_t n)
718 {
719 size_t inbuf = wxWcslen(psz) * SIZEOF_WCHAR_T;
720 size_t outbuf = n;
721 size_t res, cres;
722
723 wchar_t *tmpbuf = 0;
724
725 if (ms_wcNeedsSwap)
726 {
727 // need to copy to temp buffer to switch endianness
728 // this absolutely doesn't rock!
729 // (no, doing WC_BSWAP twice on the original buffer won't help, as it
730 // could be in read-only memory, or be accessed in some other thread)
731 tmpbuf=(wchar_t*)malloc((inbuf+1)*SIZEOF_WCHAR_T);
732 memcpy(tmpbuf,psz,(inbuf+1)*SIZEOF_WCHAR_T);
733 WC_BSWAP(tmpbuf, inbuf)
734 psz=tmpbuf;
735 }
736
737 if (buf)
738 {
739 // have destination buffer, convert there
740 cres = iconv( w2m, ICONV_CHAR_CAST(&psz), &inbuf, &buf, &outbuf );
741
742 res = n-outbuf;
743
744 // NB: iconv was given only wcslen(psz) characters on input, and so
745 // it couldn't convert the trailing zero. Let's do it ourselves
746 // if there's some room left for it in the output buffer.
747 if (res < n)
748 buf[0] = 0;
749 }
750 else
751 {
752 // no destination buffer... convert using temp buffer
753 // to calculate destination buffer requirement
754 char tbuf[16];
755 res = 0;
756 do {
757 buf = tbuf; outbuf = 16;
758
759 cres = iconv( w2m, ICONV_CHAR_CAST(&psz), &inbuf, &buf, &outbuf );
760
761 res += 16 - outbuf;
762 } while ((cres==(size_t)-1) && (errno==E2BIG));
763 }
764
765 if (ms_wcNeedsSwap)
766 {
767 free(tmpbuf);
768 }
769
770 if (ICONV_FAILED(cres, inbuf))
771 {
772 //VS: it is ok if iconv fails, hence trace only
773 wxLogTrace(wxT("strconv"), wxT("iconv failed: %s"), wxSysErrorMsg(wxSysErrorCode()));
774 return (size_t)-1;
775 }
776
777 return res;
778 }
779
780 #endif // HAVE_ICONV
781
782 // ============================================================================
783 // Win32 conversion classes
784 // ============================================================================
785
786 #if defined(__WIN32__) && !defined(__WXMICROWIN__) && !defined(__WXUNIVERSAL__)
787
788 extern long wxCharsetToCodepage(const wxChar *charset); // from utils.cpp
789
790 class CP_CharSet : public wxCharacterSet
791 {
792 public:
793 CP_CharSet(const wxChar* name)
794 : wxCharacterSet(name)
795 {
796 m_CodePage = wxCharsetToCodepage(name);
797 }
798
799 size_t MB2WC(wchar_t *buf, const char *psz, size_t n)
800 {
801 const size_t len = ::MultiByteToWideChar
802 (
803 m_CodePage, // code page
804 0, // flags (none)
805 psz, // input string
806 -1, // its length (NUL-terminated)
807 buf, // output string
808 buf ? n : 0 // size of output buffer
809 );
810
811 // note that it returns # of written chars for buf != NULL and *size*
812 // of the needed buffer for buf == NULL
813 return len ? (buf ? len : len - 1) : (size_t)-1;
814 }
815
816 size_t WC2MB(char *buf, const wchar_t *psz, size_t n)
817 {
818 const size_t len = ::WideCharToMultiByte
819 (
820 m_CodePage, // code page
821 0, // flags (none)
822 psz, // input string
823 -1, // it is (wide) NUL-terminated
824 buf, // output buffer
825 buf ? n : 0, // and its size
826 NULL, // default "replacement" char
827 NULL // [out] was it used?
828 );
829
830 // see the comment above!
831 return len ? (buf ? len : len - 1) : (size_t)-1;
832 }
833
834 bool usable() const
835 { return m_CodePage != -1; }
836
837 public:
838 long m_CodePage;
839 };
840 #endif // defined(__WIN32__) && !defined(__WXMICROWIN__) && !defined(__WXUNIVERSAL__)
841
842 // ============================================================================
843 // wxEncodingConverter based conversion classes
844 // ============================================================================
845
846 #if wxUSE_FONTMAP
847
848 class EC_CharSet : public wxCharacterSet
849 {
850 public:
851 // temporarily just use wxEncodingConverter stuff,
852 // so that it works while a better implementation is built
853 EC_CharSet(const wxChar* name) : wxCharacterSet(name),
854 enc(wxFONTENCODING_SYSTEM)
855 {
856 if (name)
857 enc = wxFontMapper::Get()->CharsetToEncoding(name, FALSE);
858
859 m_ok = m2w.Init(enc, wxFONTENCODING_UNICODE) &&
860 w2m.Init(wxFONTENCODING_UNICODE, enc);
861 }
862
863 size_t MB2WC(wchar_t *buf, const char *psz, size_t WXUNUSED(n))
864 {
865 size_t inbuf = strlen(psz);
866 if (buf)
867 m2w.Convert(psz,buf);
868 return inbuf;
869 }
870
871 size_t WC2MB(char *buf, const wchar_t *psz, size_t WXUNUSED(n))
872 {
873 const size_t inbuf = wxWcslen(psz);
874 if (buf)
875 w2m.Convert(psz,buf);
876
877 return inbuf;
878 }
879
880 bool usable() const { return m_ok; }
881
882 public:
883 wxFontEncoding enc;
884 wxEncodingConverter m2w, w2m;
885
886 // were we initialized successfully?
887 bool m_ok;
888 };
889
890 #endif // wxUSE_FONTMAP
891
892 // ----------------------------------------------------------------------------
893 // the function creating the wxCharacterSet for the specified charset on the
894 // current system, trying all possibilities
895 // ----------------------------------------------------------------------------
896
897 static wxCharacterSet *wxGetCharacterSet(const wxChar *name)
898 {
899 // check for the special case of ASCII charset
900 #if wxUSE_FONTMAP
901 if ( wxFontMapper::Get()->CharsetToEncoding(name) == wxFONTENCODING_DEFAULT )
902 #else // wxUSE_FONTMAP
903 if ( !name )
904 #endif // wxUSE_FONTMAP/!wxUSE_FONTMAP
905 {
906 // don't convert at all
907 return NULL;
908 }
909
910 // the test above must have taken care of this case
911 wxCHECK_MSG( name, NULL, _T("NULL name must be wxFONTENCODING_DEFAULT") );
912
913 wxCharacterSet *cset;
914
915 if ( wxStricmp(name, wxT("UTF8")) == 0 || wxStricmp(name, wxT("UTF-8")) == 0)
916 {
917 cset = new ID_CharSet(name, &wxConvUTF8);
918 }
919 else
920 {
921 #ifdef HAVE_ICONV
922 cset = new IC_CharSet(name);
923 #else // !HAVE_ICONV
924 cset = NULL;
925 #endif // HAVE_ICONV/!HAVE_ICONV
926 }
927
928 // it can only be NULL in this case
929 #ifndef HAVE_ICONV
930 if ( cset )
931 #endif // !HAVE_ICONV
932 {
933 if ( cset->usable() )
934 return cset;
935
936 delete cset;
937 cset = NULL;
938 }
939
940 #if defined(__WIN32__) && !defined(__WXMICROWIN__) && !defined(__WXUNIVERSAL__)
941 cset = new CP_CharSet(name);
942 if ( cset->usable() )
943 return cset;
944
945 delete cset;
946 cset = NULL;
947 #endif // defined(__WIN32__) && !defined(__WXMICROWIN__) && !defined(__WXUNIVERSAL__)
948
949 #if wxUSE_FONTMAP
950 cset = new EC_CharSet(name);
951 if ( cset->usable() )
952 return cset;
953
954 delete cset;
955 cset = NULL;
956 #endif // wxUSE_FONTMAP
957
958 wxLogError(_("Cannot convert from encoding '%s'!"), name);
959
960 return NULL;
961 }
962
963 // ============================================================================
964 // wxCSConv implementation
965 // ============================================================================
966
967 wxCSConv::wxCSConv(const wxChar *charset)
968 {
969 m_name = (wxChar *)NULL;
970 m_cset = (wxCharacterSet *) NULL;
971 m_deferred = TRUE;
972
973 SetName(charset);
974 }
975
976 wxCSConv::~wxCSConv()
977 {
978 Clear();
979 }
980
981 wxCSConv::wxCSConv(const wxCSConv& conv)
982 : wxMBConv()
983 {
984 Clear();
985 SetName(conv.m_name);
986 }
987
988 wxCSConv& wxCSConv::operator=(const wxCSConv& conv)
989 {
990 Clear();
991 SetName(conv.m_name);
992 return *this;
993 }
994
995 void wxCSConv::Clear()
996 {
997 if (m_name)
998 free(m_name);
999 if (m_cset)
1000 delete m_cset;
1001 m_name = NULL;
1002 m_cset = NULL;
1003 }
1004
1005 void wxCSConv::SetName(const wxChar *charset)
1006 {
1007 if (charset)
1008 {
1009 m_name = wxStrdup(charset);
1010 m_deferred = TRUE;
1011 }
1012 }
1013
1014 void wxCSConv::LoadNow()
1015 {
1016 if ( m_deferred )
1017 {
1018 // it would probably be better to make GetSystemEncodingName() always
1019 // available (i.e. even when wxUSE_INTL == 0)?
1020 #if wxUSE_INTL
1021 if ( !m_name )
1022 {
1023 wxString name = wxLocale::GetSystemEncodingName();
1024 if ( !name.empty() )
1025 {
1026 SetName(name);
1027 }
1028 }
1029 #endif // wxUSE_INTL
1030
1031 // wxGetCharacterSet() complains about NULL name
1032 m_cset = m_name ? wxGetCharacterSet(m_name) : NULL;
1033 m_deferred = FALSE;
1034 }
1035 }
1036
1037 size_t wxCSConv::MB2WC(wchar_t *buf, const char *psz, size_t n) const
1038 {
1039 ((wxCSConv *)this)->LoadNow(); // discard constness
1040
1041 if (m_cset)
1042 return m_cset->MB2WC(buf, psz, n);
1043
1044 // latin-1 (direct)
1045 size_t len = strlen(psz);
1046
1047 if (buf)
1048 {
1049 for (size_t c = 0; c <= len; c++)
1050 buf[c] = (unsigned char)(psz[c]);
1051 }
1052
1053 return len;
1054 }
1055
1056 size_t wxCSConv::WC2MB(char *buf, const wchar_t *psz, size_t n) const
1057 {
1058 ((wxCSConv *)this)->LoadNow(); // discard constness
1059
1060 if (m_cset)
1061 return m_cset->WC2MB(buf, psz, n);
1062
1063 // latin-1 (direct)
1064 const size_t len = wxWcslen(psz);
1065 if (buf)
1066 {
1067 for (size_t c = 0; c <= len; c++)
1068 buf[c] = (psz[c] > 0xff) ? '?' : psz[c];
1069 }
1070
1071 return len;
1072 }
1073
1074 #endif // wxUSE_WCHAR_T
1075
1076
wxWidgets (Impactor)