]> git.saurik.com Git - wxWidgets.git/blame - src/common/strconv.cpp
applied Unicode fixes for mingw32 (patch 429654)
[wxWidgets.git] / src / common / strconv.cpp
CommitLineData
6001e347
RR
1/////////////////////////////////////////////////////////////////////////////
2// Name: strconv.cpp
3// Purpose: Unicode conversion classes
4// Author: Ove Kaaven, Robert Roebling, Vadim Zeitlin
5// Modified by:
6// Created: 29/01/98
7// RCS-ID: $Id$
8// Copyright: (c) 1999 Ove Kaaven, Robert Roebling, Vadim Zeitlin
9// Licence: wxWindows license
10/////////////////////////////////////////////////////////////////////////////
11
f6bcfd97
BP
12// ============================================================================
13// declarations
14// ============================================================================
15
16// ----------------------------------------------------------------------------
17// headers
18// ----------------------------------------------------------------------------
19
6001e347
RR
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
0a1c1e62
GRG
31#ifdef __WXMSW__
32 #include "wx/msw/private.h"
33#endif
34
1cd52418 35#include <errno.h>
6001e347
RR
36#include <ctype.h>
37#include <string.h>
38#include <stdlib.h>
39
40#ifdef __SALFORDC__
41 #include <clib.h>
42#endif
43
1cd52418
OK
44#ifdef HAVE_ICONV_H
45 #include <iconv.h>
46#endif
1cd52418 47
3e61dfb0
OK
48#ifdef __WXMSW__
49 #include <windows.h>
50#endif
51
6001e347
RR
52#include "wx/debug.h"
53#include "wx/strconv.h"
3caec1bb
VS
54#include "wx/intl.h"
55#include "wx/log.h"
6001e347 56
d43088ee 57#if defined(WORDS_BIGENDIAN) || defined(__STDC_ISO_10646__)
1cd52418
OK
58#define BSWAP_UCS4(str, len)
59#define BSWAP_UCS2(str, len)
60#else
61#define BSWAP_UCS4(str, len) { unsigned _c; for (_c=0; _c<len; _c++) str[_c]=wxUINT32_SWAP_ALWAYS(str[_c]); }
62#define BSWAP_UCS2(str, len) { unsigned _c; for (_c=0; _c<len; _c++) str[_c]=wxUINT16_SWAP_ALWAYS(str[_c]); }
63#define WC_NEED_BSWAP
64#endif
65#define BSWAP_UTF32(str, len) BSWAP_UCS4(str, len)
66#define BSWAP_UTF16(str, len) BSWAP_UCS2(str, len)
67
68#if SIZEOF_WCHAR_T == 4
69#define WC_NAME "UCS4"
70#define WC_BSWAP BSWAP_UCS4
71#elif SIZEOF_WCHAR_T == 2
72#define WC_NAME "UTF16"
73#define WC_BSWAP BSWAP_UTF16
74#define WC_UTF16
75#endif
76
f6bcfd97
BP
77// ----------------------------------------------------------------------------
78// globals
79// ----------------------------------------------------------------------------
6001e347
RR
80
81WXDLLEXPORT_DATA(wxMBConv *) wxConvCurrent = &wxConvLibc;
82
f6bcfd97
BP
83// ============================================================================
84// implementation
85// ============================================================================
6001e347 86
f6bcfd97 87#if wxUSE_WCHAR_T
6001e347 88
b0a6bb75
VZ
89#ifdef WC_UTF16
90
1cd52418
OK
91static size_t encode_utf16(wxUint32 input,wxUint16*output)
92{
dccce9ea 93 if (input<=0xffff)
4def3b35
VS
94 {
95 if (output) *output++ = input;
96 return 1;
dccce9ea
VZ
97 }
98 else if (input>=0x110000)
4def3b35
VS
99 {
100 return (size_t)-1;
dccce9ea
VZ
101 }
102 else
4def3b35 103 {
dccce9ea 104 if (output)
4def3b35
VS
105 {
106 *output++ = (input >> 10)+0xd7c0;
107 *output++ = (input&0x3ff)+0xdc00;
108 }
109 return 2;
1cd52418 110 }
1cd52418
OK
111}
112
113static size_t decode_utf16(wxUint16*input,wxUint32&output)
114{
dccce9ea 115 if ((*input<0xd800) || (*input>0xdfff))
4def3b35
VS
116 {
117 output = *input;
118 return 1;
dccce9ea
VZ
119 }
120 else if ((input[1]<0xdc00) || (input[1]>=0xdfff))
4def3b35
VS
121 {
122 output = *input;
123 return (size_t)-1;
dccce9ea
VZ
124 }
125 else
4def3b35
VS
126 {
127 output = ((input[0] - 0xd7c0) << 10) + (input[1] - 0xdc00);
128 return 2;
129 }
1cd52418
OK
130}
131
b0a6bb75
VZ
132#endif // WC_UTF16
133
f6bcfd97 134// ----------------------------------------------------------------------------
6001e347 135// wxMBConv
f6bcfd97 136// ----------------------------------------------------------------------------
6001e347
RR
137
138WXDLLEXPORT_DATA(wxMBConv) wxConvLibc;
139
140size_t wxMBConv::MB2WC(wchar_t *buf, const char *psz, size_t n) const
141{
142 return wxMB2WC(buf, psz, n);
143}
144
145size_t wxMBConv::WC2MB(char *buf, const wchar_t *psz, size_t n) const
146{
147 return wxWC2MB(buf, psz, n);
148}
149
150const wxWCharBuffer wxMBConv::cMB2WC(const char *psz) const
151{
f6bcfd97 152 if (psz)
6001e347
RR
153 {
154 size_t nLen = MB2WC((wchar_t *) NULL, psz, 0);
f6bcfd97
BP
155 if (nLen == (size_t)-1)
156 return wxWCharBuffer((wchar_t *) NULL);
6001e347
RR
157 wxWCharBuffer buf(nLen);
158 MB2WC((wchar_t *)(const wchar_t *) buf, psz, nLen);
159 return buf;
f6bcfd97
BP
160 }
161 else
6001e347
RR
162 return wxWCharBuffer((wchar_t *) NULL);
163}
164
165const wxCharBuffer wxMBConv::cWC2MB(const wchar_t *psz) const
166{
f6bcfd97 167 if (psz)
6001e347
RR
168 {
169 size_t nLen = WC2MB((char *) NULL, psz, 0);
f6bcfd97
BP
170 if (nLen == (size_t)-1)
171 return wxCharBuffer((char *) NULL);
6001e347
RR
172 wxCharBuffer buf(nLen);
173 WC2MB((char *)(const char *) buf, psz, nLen);
174 return buf;
f6bcfd97
BP
175 }
176 else
6001e347
RR
177 return wxCharBuffer((char *) NULL);
178}
179
f6bcfd97 180// ----------------------------------------------------------------------------
6001e347 181// standard file conversion
f6bcfd97 182// ----------------------------------------------------------------------------
6001e347
RR
183
184WXDLLEXPORT_DATA(wxMBConvFile) wxConvFile;
185
186// just use the libc conversion for now
187size_t wxMBConvFile::MB2WC(wchar_t *buf, const char *psz, size_t n) const
188{
189 return wxMB2WC(buf, psz, n);
190}
191
192size_t wxMBConvFile::WC2MB(char *buf, const wchar_t *psz, size_t n) const
193{
194 return wxWC2MB(buf, psz, n);
195}
196
f6bcfd97 197// ----------------------------------------------------------------------------
6001e347 198// standard gdk conversion
f6bcfd97
BP
199// ----------------------------------------------------------------------------
200
201#ifdef __WXGTK12__
6001e347
RR
202
203WXDLLEXPORT_DATA(wxMBConvGdk) wxConvGdk;
204
205#include <gdk/gdk.h>
206
207size_t wxMBConvGdk::MB2WC(wchar_t *buf, const char *psz, size_t n) const
208{
dccce9ea 209 if (buf)
4def3b35
VS
210 {
211 return gdk_mbstowcs((GdkWChar *)buf, psz, n);
dccce9ea
VZ
212 }
213 else
4def3b35
VS
214 {
215 GdkWChar *nbuf = new GdkWChar[n=strlen(psz)];
216 size_t len = gdk_mbstowcs(nbuf, psz, n);
217 delete[] nbuf;
218 return len;
219 }
6001e347
RR
220}
221
222size_t wxMBConvGdk::WC2MB(char *buf, const wchar_t *psz, size_t n) const
223{
4def3b35
VS
224 char *mbstr = gdk_wcstombs((GdkWChar *)psz);
225 size_t len = mbstr ? strlen(mbstr) : 0;
dccce9ea 226 if (buf)
4def3b35 227 {
dccce9ea 228 if (len > n)
4def3b35
VS
229 len = n;
230 memcpy(buf, psz, len);
dccce9ea 231 if (len < n)
4def3b35
VS
232 buf[len] = 0;
233 }
234 return len;
6001e347 235}
f6bcfd97 236
6001e347
RR
237#endif // GTK > 1.0
238
239// ----------------------------------------------------------------------------
240// UTF-7
241// ----------------------------------------------------------------------------
242
243WXDLLEXPORT_DATA(wxMBConvUTF7) wxConvUTF7;
244
245#if 0
246static char utf7_setD[]="ABCDEFGHIJKLMNOPQRSTUVWXYZ"
247 "abcdefghijklmnopqrstuvwxyz"
248 "0123456789'(),-./:?";
249static char utf7_setO[]="!\"#$%&*;<=>@[]^_`{|}";
250static char utf7_setB[]="ABCDEFGHIJKLMNOPQRSTUVWXYZ"
251 "abcdefghijklmnopqrstuvwxyz"
252 "0123456789+/";
253#endif
254
255// TODO: write actual implementations of UTF-7 here
256size_t wxMBConvUTF7::MB2WC(wchar_t * WXUNUSED(buf),
257 const char * WXUNUSED(psz),
258 size_t WXUNUSED(n)) const
259{
260 return 0;
261}
262
263size_t wxMBConvUTF7::WC2MB(char * WXUNUSED(buf),
264 const wchar_t * WXUNUSED(psz),
265 size_t WXUNUSED(n)) const
266{
267 return 0;
268}
269
f6bcfd97 270// ----------------------------------------------------------------------------
6001e347 271// UTF-8
f6bcfd97 272// ----------------------------------------------------------------------------
6001e347
RR
273
274WXDLLEXPORT_DATA(wxMBConvUTF8) wxConvUTF8;
275
dccce9ea 276static wxUint32 utf8_max[]=
4def3b35 277 { 0x7f, 0x7ff, 0xffff, 0x1fffff, 0x3ffffff, 0x7fffffff, 0xffffffff };
6001e347
RR
278
279size_t wxMBConvUTF8::MB2WC(wchar_t *buf, const char *psz, size_t n) const
280{
4def3b35
VS
281 size_t len = 0;
282
dccce9ea 283 while (*psz && ((!buf) || (len < n)))
4def3b35
VS
284 {
285 unsigned char cc = *psz++, fc = cc;
286 unsigned cnt;
dccce9ea 287 for (cnt = 0; fc & 0x80; cnt++)
4def3b35 288 fc <<= 1;
dccce9ea 289 if (!cnt)
4def3b35
VS
290 {
291 // plain ASCII char
dccce9ea 292 if (buf)
4def3b35
VS
293 *buf++ = cc;
294 len++;
dccce9ea
VZ
295 }
296 else
4def3b35
VS
297 {
298 cnt--;
dccce9ea 299 if (!cnt)
4def3b35
VS
300 {
301 // invalid UTF-8 sequence
302 return (size_t)-1;
dccce9ea
VZ
303 }
304 else
4def3b35
VS
305 {
306 unsigned ocnt = cnt - 1;
307 wxUint32 res = cc & (0x3f >> cnt);
dccce9ea 308 while (cnt--)
4def3b35
VS
309 {
310 cc = *psz++;
dccce9ea 311 if ((cc & 0xC0) != 0x80)
4def3b35
VS
312 {
313 // invalid UTF-8 sequence
314 return (size_t)-1;
315 }
316 res = (res << 6) | (cc & 0x3f);
317 }
dccce9ea 318 if (res <= utf8_max[ocnt])
4def3b35
VS
319 {
320 // illegal UTF-8 encoding
321 return (size_t)-1;
322 }
1cd52418 323#ifdef WC_UTF16
4def3b35
VS
324 size_t pa = encode_utf16(res, buf);
325 if (pa == (size_t)-1)
326 return (size_t)-1;
dccce9ea 327 if (buf)
4def3b35
VS
328 buf += pa;
329 len += pa;
1cd52418 330#else
dccce9ea 331 if (buf)
4def3b35
VS
332 *buf++ = res;
333 len++;
1cd52418 334#endif
4def3b35
VS
335 }
336 }
6001e347 337 }
dccce9ea 338 if (buf && (len < n))
4def3b35
VS
339 *buf = 0;
340 return len;
6001e347
RR
341}
342
343size_t wxMBConvUTF8::WC2MB(char *buf, const wchar_t *psz, size_t n) const
344{
4def3b35 345 size_t len = 0;
6001e347 346
dccce9ea 347 while (*psz && ((!buf) || (len < n)))
4def3b35
VS
348 {
349 wxUint32 cc;
1cd52418 350#ifdef WC_UTF16
4def3b35
VS
351 size_t pa = decode_utf16(psz,cc);
352 psz += (pa == (size_t)-1) ? 1 : pa;
1cd52418 353#else
4def3b35
VS
354 cc=(*psz++) & 0x7fffffff;
355#endif
356 unsigned cnt;
357 for (cnt = 0; cc > utf8_max[cnt]; cnt++) {}
dccce9ea 358 if (!cnt)
4def3b35
VS
359 {
360 // plain ASCII char
dccce9ea 361 if (buf)
4def3b35
VS
362 *buf++ = cc;
363 len++;
dccce9ea
VZ
364 }
365
366 else
4def3b35
VS
367 {
368 len += cnt + 1;
dccce9ea 369 if (buf)
4def3b35
VS
370 {
371 *buf++ = (-128 >> cnt) | ((cc >> (cnt * 6)) & (0x3f >> cnt));
372 while (cnt--)
373 *buf++ = 0x80 | ((cc >> (cnt * 6)) & 0x3f);
374 }
375 }
6001e347 376 }
4def3b35
VS
377
378 if (buf && (len<n)) *buf = 0;
379 return len;
6001e347
RR
380}
381
382// ----------------------------------------------------------------------------
383// specified character set
384// ----------------------------------------------------------------------------
385
f6bcfd97
BP
386WXDLLEXPORT_DATA(wxCSConv) wxConvLocal((const wxChar *)NULL);
387
388#include "wx/encconv.h"
389#include "wx/fontmap.h"
6001e347 390
1cd52418
OK
391// TODO: add some tables here
392// - perhaps common encodings to common codepages (for Win32)
393// - perhaps common encodings to objects ("UTF8" -> wxConvUTF8)
394// - move wxEncodingConverter meat in here
395
396#ifdef __WIN32__
397#include "wx/msw/registry.h"
398// this should work if M$ Internet Exploiter is installed
399static long CharsetToCodepage(const wxChar *name)
400{
dccce9ea 401 if (!name)
f1339c56 402 return GetACP();
dccce9ea 403
f1339c56 404 long CP=-1;
dccce9ea 405
f1339c56
RR
406 wxString cn(name);
407 do {
5ce0e4ac 408 wxString path(wxT("MIME\\Database\\Charset\\"));
f1339c56 409 path += cn;
5ce0e4ac
VS
410 wxRegKey key(wxRegKey::HKCR, path);
411
412 if (!key.Exists()) continue;
dccce9ea 413
5ce0e4ac
VS
414 // two cases: either there's an AliasForCharset string,
415 // or there are Codepage and InternetEncoding dwords.
416 // The InternetEncoding gives us the actual encoding,
417 // the Codepage just says which Windows character set to
418 // use when displaying the data.
419 if (key.HasValue(wxT("InternetEncoding")) &&
420 key.QueryValue(wxT("InternetEncoding"), &CP)) break;
dccce9ea 421
f1339c56 422 // no encoding, see if it's an alias
5ce0e4ac
VS
423 if (!key.HasValue(wxT("AliasForCharset")) ||
424 !key.QueryValue(wxT("AliasForCharset"), cn)) break;
f1339c56 425 } while (1);
dccce9ea 426
f1339c56 427 return CP;
1cd52418
OK
428}
429#endif
430
6001e347
RR
431class wxCharacterSet
432{
1cd52418 433public:
f1339c56
RR
434 wxCharacterSet(const wxChar*name)
435 : cname(name) {}
436 virtual ~wxCharacterSet()
437 {}
dccce9ea 438 virtual size_t MB2WC(wchar_t *buf, const char *psz, size_t n)
f1339c56 439 { return (size_t)-1; }
4def3b35 440 virtual size_t WC2MB(char *buf, const wchar_t *psz, size_t n)
f1339c56
RR
441 { return (size_t)-1; }
442 virtual bool usable()
443 { return FALSE; }
444public:
445 const wxChar*cname;
1cd52418
OK
446};
447
448class ID_CharSet : public wxCharacterSet
449{
450public:
4def3b35 451 ID_CharSet(const wxChar *name,wxMBConv *cnv)
f1339c56 452 : wxCharacterSet(name), work(cnv) {}
dccce9ea 453
4def3b35 454 size_t MB2WC(wchar_t *buf, const char *psz, size_t n)
f1339c56 455 { return work ? work->MB2WC(buf,psz,n) : (size_t)-1; }
dccce9ea 456
4def3b35 457 size_t WC2MB(char *buf, const wchar_t *psz, size_t n)
f1339c56 458 { return work ? work->WC2MB(buf,psz,n) : (size_t)-1; }
dccce9ea 459
f1339c56
RR
460 bool usable()
461 { return work!=NULL; }
462public:
463 wxMBConv*work;
1cd52418
OK
464};
465
3caec1bb 466
1cd52418 467#ifdef HAVE_ICONV_H
3caec1bb
VS
468
469// VS: glibc 2.1.3 is broken in that iconv() conversion to/from UCS4 fails with E2BIG
470// if output buffer is _exactly_ as big as needed. Such case is (unless there's
471// yet another bug in glibc) the only case when iconv() returns with (size_t)-1
472// (which means error) and says there are 0 bytes left in the input buffer --
473// when _real_ error occurs, bytes-left-in-input buffer is non-zero. Hence,
474// this alternative test for iconv() failure.
475// [This bug does not appear in glibc 2.2.]
476#if defined(__GLIBC__) && __GLIBC__ == 2 && __GLIBC_MINOR__ <= 1
477#define ICONV_FAILED(cres, bufLeft) ((cres == (size_t)-1) && \
478 (errno != E2BIG || bufLeft != 0))
479#else
480#define ICONV_FAILED(cres, bufLeft) (cres == (size_t)-1)
481#endif
482
1cd52418
OK
483class IC_CharSet : public wxCharacterSet
484{
485public:
dccce9ea 486 IC_CharSet(const wxChar *name)
3caec1bb 487 : wxCharacterSet(name)
f1339c56 488 {
3caec1bb
VS
489 m2w = iconv_open(WC_NAME, wxConvLibc.cWX2MB(cname));
490 w2m = iconv_open(wxConvLibc.cWX2MB(cname), WC_NAME);
f1339c56 491 }
dccce9ea 492
3caec1bb
VS
493 ~IC_CharSet()
494 {
dccce9ea 495 if ( m2w != (iconv_t)-1 )
3caec1bb 496 iconv_close(m2w);
dccce9ea 497 if ( w2m != (iconv_t)-1 )
3caec1bb 498 iconv_close(w2m);
f1339c56 499 }
dccce9ea 500
3caec1bb 501 size_t MB2WC(wchar_t *buf, const char *psz, size_t n)
f1339c56 502 {
f1339c56 503 size_t inbuf = strlen(psz);
3caec1bb 504 size_t outbuf = n * SIZEOF_WCHAR_T;
f1339c56 505 size_t res, cres;
3caec1bb
VS
506 // VS: Use these instead of psz, buf because iconv() modifies its arguments:
507 wchar_t *bufPtr = buf;
508 const char *pszPtr = psz;
509
f1339c56
RR
510 if (buf)
511 {
512 // have destination buffer, convert there
95c8801c 513#ifdef WX_ICONV_TAKES_CHAR
3caec1bb 514 cres = iconv(m2w, (char**)&pszPtr, &inbuf, (char**)&bufPtr, &outbuf);
95c8801c 515#else
3caec1bb 516 cres = iconv(m2w, &pszPtr, &inbuf, (char**)&bufPtr, &outbuf);
95c8801c 517#endif
3caec1bb 518 res = n - (outbuf / SIZEOF_WCHAR_T);
f1339c56 519 // convert to native endianness
3caec1bb
VS
520#ifdef WC_NEED_BSWAP
521 WC_BSWAP(buf /* _not_ bufPtr */, res)
522#endif
f1339c56
RR
523 }
524 else
525 {
526 // no destination buffer... convert using temp buffer
527 // to calculate destination buffer requirement
528 wchar_t tbuf[8];
529 res = 0;
530 do {
3caec1bb 531 bufPtr = tbuf; outbuf = 8*SIZEOF_WCHAR_T;
95c8801c 532#ifdef WX_ICONV_TAKES_CHAR
3caec1bb 533 cres = iconv( m2w, (char**)&pszPtr, &inbuf, (char**)&bufPtr, &outbuf );
95c8801c 534#else
3caec1bb 535 cres = iconv( m2w, &pszPtr, &inbuf, (char**)&bufPtr, &outbuf );
95c8801c 536#endif
f1339c56
RR
537 res += 8-(outbuf/SIZEOF_WCHAR_T);
538 } while ((cres==(size_t)-1) && (errno==E2BIG));
539 }
dccce9ea 540
3caec1bb 541 if (ICONV_FAILED(cres, inbuf))
f1339c56 542 return (size_t)-1;
3caec1bb 543
f1339c56
RR
544 return res;
545 }
dccce9ea 546
4def3b35 547 size_t WC2MB(char *buf, const wchar_t *psz, size_t n)
f1339c56 548 {
1cd52418 549#if defined(__BORLANDC__) && (__BORLANDC__ > 0x530)
3caec1bb 550 size_t inbuf = std::wcslen(psz) * SIZEOF_WCHAR_T;
1cd52418 551#else
3caec1bb 552 size_t inbuf = ::wcslen(psz) * SIZEOF_WCHAR_T;
1cd52418 553#endif
f1339c56
RR
554 size_t outbuf = n;
555 size_t res, cres;
3caec1bb 556
1cd52418 557#ifdef WC_NEED_BSWAP
f1339c56
RR
558 // need to copy to temp buffer to switch endianness
559 // this absolutely doesn't rock!
560 // (no, doing WC_BSWAP twice on the original buffer won't help, as it
561 // could be in read-only memory, or be accessed in some other thread)
4def3b35 562 wchar_t *tmpbuf=(wchar_t*)malloc((inbuf+1)*SIZEOF_WCHAR_T);
f1339c56
RR
563 memcpy(tmpbuf,psz,(inbuf+1)*SIZEOF_WCHAR_T);
564 WC_BSWAP(tmpbuf, inbuf)
565 psz=tmpbuf;
1cd52418 566#endif
f1339c56
RR
567 if (buf)
568 {
569 // have destination buffer, convert there
95c8801c 570#ifdef WX_ICONV_TAKES_CHAR
f1339c56 571 cres = iconv( w2m, (char**)&psz, &inbuf, &buf, &outbuf );
95c8801c
VS
572#else
573 cres = iconv( w2m, (const char**)&psz, &inbuf, &buf, &outbuf );
574#endif
f1339c56
RR
575 res = n-outbuf;
576 }
577 else
578 {
579 // no destination buffer... convert using temp buffer
580 // to calculate destination buffer requirement
581 char tbuf[16];
582 res = 0;
583 do {
584 buf = tbuf; outbuf = 16;
95c8801c 585#ifdef WX_ICONV_TAKES_CHAR
f1339c56 586 cres = iconv( w2m, (char**)&psz, &inbuf, &buf, &outbuf );
95c8801c
VS
587#else
588 cres = iconv( w2m, (const char**)&psz, &inbuf, &buf, &outbuf );
589#endif
f1339c56
RR
590 res += 16 - outbuf;
591 } while ((cres==(size_t)-1) && (errno==E2BIG));
592 }
1cd52418 593#ifdef WC_NEED_BSWAP
f1339c56 594 free(tmpbuf);
1cd52418 595#endif
3caec1bb 596 if (ICONV_FAILED(cres, inbuf))
f1339c56 597 return (size_t)-1;
dccce9ea 598
f1339c56
RR
599 return res;
600 }
dccce9ea 601
f1339c56 602 bool usable()
3caec1bb 603 { return (m2w != (iconv_t)-1) && (w2m != (iconv_t)-1); }
dccce9ea 604
f1339c56
RR
605public:
606 iconv_t m2w, w2m;
1cd52418
OK
607};
608#endif
609
610#ifdef __WIN32__
611class CP_CharSet : public wxCharacterSet
612{
613public:
dccce9ea 614 CP_CharSet(const wxChar*name)
f1339c56 615 : wxCharacterSet(name), CodePage(CharsetToCodepage(name)) {}
dccce9ea 616
4def3b35 617 size_t MB2WC(wchar_t *buf, const char *psz, size_t n)
f1339c56 618 {
dccce9ea 619 size_t len =
4def3b35 620 MultiByteToWideChar(CodePage, 0, psz, -1, buf, buf ? n : 0);
f1339c56
RR
621 return len ? len : (size_t)-1;
622 }
dccce9ea 623
4def3b35 624 size_t WC2MB(char *buf, const wchar_t *psz, size_t n)
f1339c56 625 {
4def3b35
VS
626 size_t len = WideCharToMultiByte(CodePage, 0, psz, -1, buf,
627 buf ? n : 0, NULL, NULL);
f1339c56
RR
628 return len ? len : (size_t)-1;
629 }
dccce9ea 630
f1339c56 631 bool usable()
4def3b35 632 { return CodePage != -1; }
f1339c56
RR
633
634public:
635 long CodePage;
1cd52418
OK
636};
637#endif
638
639class EC_CharSet : public wxCharacterSet
640{
6001e347 641public:
f1339c56
RR
642 // temporarily just use wxEncodingConverter stuff,
643 // so that it works while a better implementation is built
dccce9ea 644 EC_CharSet(const wxChar*name) : wxCharacterSet(name),
4def3b35 645 enc(wxFONTENCODING_SYSTEM)
f1339c56
RR
646 {
647 if (name)
648 enc = wxTheFontMapper->CharsetToEncoding(name, FALSE);
649 m2w.Init(enc, wxFONTENCODING_UNICODE);
650 w2m.Init(wxFONTENCODING_UNICODE, enc);
651 }
dccce9ea 652
4def3b35 653 size_t MB2WC(wchar_t *buf, const char *psz, size_t n)
f1339c56
RR
654 {
655 size_t inbuf = strlen(psz);
dccce9ea 656 if (buf)
4def3b35 657 m2w.Convert(psz,buf);
f1339c56
RR
658 return inbuf;
659 }
dccce9ea 660
4def3b35 661 size_t WC2MB(char *buf, const wchar_t *psz, size_t n)
f1339c56 662 {
1cd52418 663#if defined(__BORLANDC__) && (__BORLANDC__ > 0x530)
f1339c56 664 size_t inbuf = std::wcslen(psz);
1cd52418 665#else
f1339c56 666 size_t inbuf = ::wcslen(psz);
1cd52418 667#endif
f1339c56
RR
668 if (buf)
669 w2m.Convert(psz,buf);
dccce9ea 670
f1339c56
RR
671 return inbuf;
672 }
dccce9ea 673
f1339c56
RR
674 bool usable()
675 { return (enc!=wxFONTENCODING_SYSTEM) && (enc!=wxFONTENCODING_DEFAULT); }
676
677public:
678 wxFontEncoding enc;
679 wxEncodingConverter m2w, w2m;
f6bcfd97 680};
6001e347 681
f6bcfd97 682static wxCharacterSet *wxGetCharacterSet(const wxChar *name)
6001e347 683{
f1339c56
RR
684 wxCharacterSet *cset = NULL;
685 if (name)
686 {
4def3b35 687 if (wxStricmp(name, wxT("UTF8")) == 0 || wxStricmp(name, wxT("UTF-8")) == 0)
f1339c56
RR
688 {
689 cset = new ID_CharSet(name, &wxConvUTF8);
690 }
691 else
692 {
1cd52418 693#ifdef HAVE_ICONV_H
f1339c56 694 cset = new IC_CharSet(name); // may not take NULL
1cd52418 695#endif
f1339c56 696 }
1cd52418 697 }
dccce9ea 698
f1339c56 699 if (cset && cset->usable()) return cset;
dccce9ea
VZ
700 if (cset)
701 {
702 delete cset;
703 cset = NULL;
704 }
705
1cd52418 706#ifdef __WIN32__
f1339c56 707 cset = new CP_CharSet(name); // may take NULL
dccce9ea
VZ
708 if (cset->usable())
709 return cset;
710
711 delete cset;
712#endif // __WIN32__
713
f1339c56 714 cset = new EC_CharSet(name);
dccce9ea
VZ
715 if (cset->usable())
716 return cset;
717
f1339c56 718 delete cset;
3caec1bb 719 wxLogError(_("Unknown encoding '%s'!"), name);
f1339c56 720 return NULL;
6001e347
RR
721}
722
6001e347
RR
723wxCSConv::wxCSConv(const wxChar *charset)
724{
dccce9ea 725 m_name = (wxChar *)NULL;
f1339c56 726 m_cset = (wxCharacterSet *) NULL;
82713003
VZ
727 m_deferred = TRUE;
728
f1339c56 729 SetName(charset);
6001e347
RR
730}
731
732wxCSConv::~wxCSConv()
733{
dccce9ea
VZ
734 free(m_name);
735 delete m_cset;
6001e347
RR
736}
737
738void wxCSConv::SetName(const wxChar *charset)
739{
f1339c56
RR
740 if (charset)
741 {
742 m_name = wxStrdup(charset);
743 m_deferred = TRUE;
744 }
6001e347
RR
745}
746
747void wxCSConv::LoadNow()
748{
f1339c56
RR
749 if (m_deferred)
750 {
dccce9ea 751 if ( !m_name )
f1339c56 752 {
dccce9ea
VZ
753 wxString name = wxLocale::GetSystemEncodingName();
754 if ( !name.empty() )
755 SetName(name);
f1339c56 756 }
dccce9ea 757
f1339c56
RR
758 m_cset = wxGetCharacterSet(m_name);
759 m_deferred = FALSE;
6001e347 760 }
6001e347
RR
761}
762
763size_t wxCSConv::MB2WC(wchar_t *buf, const char *psz, size_t n) const
764{
f1339c56 765 ((wxCSConv *)this)->LoadNow(); // discard constness
dccce9ea 766
f1339c56
RR
767 if (m_cset)
768 return m_cset->MB2WC(buf, psz, n);
769
770 // latin-1 (direct)
4def3b35 771 size_t len = strlen(psz);
dccce9ea 772
f1339c56
RR
773 if (buf)
774 {
4def3b35 775 for (size_t c = 0; c <= len; c++)
f1339c56
RR
776 buf[c] = (unsigned char)(psz[c]);
777 }
dccce9ea 778
f1339c56 779 return len;
6001e347
RR
780}
781
782size_t wxCSConv::WC2MB(char *buf, const wchar_t *psz, size_t n) const
783{
f1339c56 784 ((wxCSConv *)this)->LoadNow(); // discard constness
dccce9ea 785
f1339c56
RR
786 if (m_cset)
787 return m_cset->WC2MB(buf, psz, n);
1cd52418 788
f1339c56 789 // latin-1 (direct)
d834f22c 790#if defined(__BORLANDC__) && (__BORLANDC__ > 0x530)
f1339c56 791 size_t len=std::wcslen(psz);
d834f22c 792#else
f1339c56 793 size_t len=::wcslen(psz);
d834f22c 794#endif
f1339c56
RR
795 if (buf)
796 {
4def3b35
VS
797 for (size_t c = 0; c <= len; c++)
798 buf[c] = (psz[c] > 0xff) ? '?' : psz[c];
f1339c56 799 }
dccce9ea 800
f1339c56 801 return len;
6001e347
RR
802}
803
1cd52418 804#ifdef HAVE_ICONV_H
dccce9ea 805
1cd52418
OK
806class IC_CharSetConverter
807{
808public:
4def3b35
VS
809 IC_CharSetConverter(IC_CharSet *from, IC_CharSet *to)
810 {
dccce9ea
VZ
811 cnv = iconv_open(wxConvLibc.cWX2MB(to->cname),
812 wxConvLibc.cWX2MB(from->cname));
4def3b35 813 }
dccce9ea 814
f1339c56 815 ~IC_CharSetConverter()
dccce9ea
VZ
816 {
817 if (cnv != (iconv_t)-1)
818 iconv_close(cnv);
4def3b35 819 }
dccce9ea 820
4def3b35 821 size_t Convert(char *buf, const char *psz, size_t n)
f1339c56
RR
822 {
823 size_t inbuf = strlen(psz);
824 size_t outbuf = n;
95c8801c 825#ifdef WX_ICONV_TAKES_CHAR
f1339c56 826 size_t res = iconv( cnv, (char**)&psz, &inbuf, &buf, &outbuf );
95c8801c
VS
827#else
828 size_t res = iconv( cnv, &psz, &inbuf, &buf, &outbuf );
829#endif
dccce9ea 830 if (res == (size_t)-1)
4def3b35
VS
831 return (size_t)-1;
832 return (n - outbuf);
f1339c56
RR
833 }
834
835public:
836 iconv_t cnv;
1cd52418 837};
dccce9ea
VZ
838
839#endif // HAVE_ICONV_H
1cd52418
OK
840
841class EC_CharSetConverter
842{
843public:
f1339c56
RR
844 EC_CharSetConverter(EC_CharSet*from,EC_CharSet*to)
845 { cnv.Init(from->enc,to->enc); }
dccce9ea 846
f1339c56
RR
847 size_t Convert(char*buf, const char*psz, size_t n)
848 {
849 size_t inbuf = strlen(psz);
850 if (buf) cnv.Convert(psz,buf);
851 return inbuf;
852 }
dccce9ea 853
f1339c56
RR
854public:
855 wxEncodingConverter cnv;
1cd52418
OK
856};
857
f6bcfd97
BP
858#else // !wxUSE_WCHAR_T
859
860// ----------------------------------------------------------------------------
861// stand-ins in absence of wchar_t
862// ----------------------------------------------------------------------------
863
864WXDLLEXPORT_DATA(wxMBConv) wxConvLibc, wxConvFile;
865
866#endif // wxUSE_WCHAR_T
6001e347
RR
867
868