]>
Commit | Line | Data |
---|---|---|
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 | ||
a3f2769e VZ |
68 | // under Unix SIZEOF_WCHAR_T is defined by configure, but under other platforms |
69 | // it might be not defined - assume the most common value | |
70 | #ifndef SIZEOF_WCHAR_T | |
71 | #define SIZEOF_WCHAR_T 2 | |
72 | #endif // !defined(SIZEOF_WCHAR_T) | |
73 | ||
1cd52418 | 74 | #if SIZEOF_WCHAR_T == 4 |
a3f2769e VZ |
75 | #define WC_NAME "UCS4" |
76 | #define WC_BSWAP BSWAP_UCS4 | |
1cd52418 | 77 | #elif SIZEOF_WCHAR_T == 2 |
a3f2769e VZ |
78 | #define WC_NAME "UTF16" |
79 | #define WC_BSWAP BSWAP_UTF16 | |
80 | #define WC_UTF16 | |
bab1e722 | 81 | #else // sizeof(wchar_t) != 2 nor 4 |
a3f2769e VZ |
82 | // I don't know what to do about this |
83 | #error "Weird sizeof(wchar_t): please report your platform details to wx-users mailing list" | |
1cd52418 OK |
84 | #endif |
85 | ||
f6bcfd97 BP |
86 | // ---------------------------------------------------------------------------- |
87 | // globals | |
88 | // ---------------------------------------------------------------------------- | |
6001e347 RR |
89 | |
90 | WXDLLEXPORT_DATA(wxMBConv *) wxConvCurrent = &wxConvLibc; | |
91 | ||
f6bcfd97 BP |
92 | // ============================================================================ |
93 | // implementation | |
94 | // ============================================================================ | |
6001e347 | 95 | |
f6bcfd97 | 96 | #if wxUSE_WCHAR_T |
6001e347 | 97 | |
b0a6bb75 VZ |
98 | #ifdef WC_UTF16 |
99 | ||
eccf1b2c | 100 | static size_t encode_utf16(wxUint32 input, wchar_t *output) |
1cd52418 | 101 | { |
dccce9ea | 102 | if (input<=0xffff) |
4def3b35 VS |
103 | { |
104 | if (output) *output++ = input; | |
105 | return 1; | |
dccce9ea VZ |
106 | } |
107 | else if (input>=0x110000) | |
4def3b35 VS |
108 | { |
109 | return (size_t)-1; | |
dccce9ea VZ |
110 | } |
111 | else | |
4def3b35 | 112 | { |
dccce9ea | 113 | if (output) |
4def3b35 VS |
114 | { |
115 | *output++ = (input >> 10)+0xd7c0; | |
116 | *output++ = (input&0x3ff)+0xdc00; | |
117 | } | |
118 | return 2; | |
1cd52418 | 119 | } |
1cd52418 OK |
120 | } |
121 | ||
eccf1b2c | 122 | static size_t decode_utf16(const wchar_t* input, wxUint32& output) |
1cd52418 | 123 | { |
dccce9ea | 124 | if ((*input<0xd800) || (*input>0xdfff)) |
4def3b35 VS |
125 | { |
126 | output = *input; | |
127 | return 1; | |
dccce9ea VZ |
128 | } |
129 | else if ((input[1]<0xdc00) || (input[1]>=0xdfff)) | |
4def3b35 VS |
130 | { |
131 | output = *input; | |
132 | return (size_t)-1; | |
dccce9ea VZ |
133 | } |
134 | else | |
4def3b35 VS |
135 | { |
136 | output = ((input[0] - 0xd7c0) << 10) + (input[1] - 0xdc00); | |
137 | return 2; | |
138 | } | |
1cd52418 OK |
139 | } |
140 | ||
b0a6bb75 VZ |
141 | #endif // WC_UTF16 |
142 | ||
f6bcfd97 | 143 | // ---------------------------------------------------------------------------- |
6001e347 | 144 | // wxMBConv |
f6bcfd97 | 145 | // ---------------------------------------------------------------------------- |
6001e347 RR |
146 | |
147 | WXDLLEXPORT_DATA(wxMBConv) wxConvLibc; | |
148 | ||
149 | size_t wxMBConv::MB2WC(wchar_t *buf, const char *psz, size_t n) const | |
150 | { | |
151 | return wxMB2WC(buf, psz, n); | |
152 | } | |
153 | ||
154 | size_t wxMBConv::WC2MB(char *buf, const wchar_t *psz, size_t n) const | |
155 | { | |
156 | return wxWC2MB(buf, psz, n); | |
157 | } | |
158 | ||
159 | const wxWCharBuffer wxMBConv::cMB2WC(const char *psz) const | |
160 | { | |
f6bcfd97 | 161 | if (psz) |
6001e347 RR |
162 | { |
163 | size_t nLen = MB2WC((wchar_t *) NULL, psz, 0); | |
f6bcfd97 BP |
164 | if (nLen == (size_t)-1) |
165 | return wxWCharBuffer((wchar_t *) NULL); | |
6001e347 RR |
166 | wxWCharBuffer buf(nLen); |
167 | MB2WC((wchar_t *)(const wchar_t *) buf, psz, nLen); | |
168 | return buf; | |
f6bcfd97 BP |
169 | } |
170 | else | |
6001e347 RR |
171 | return wxWCharBuffer((wchar_t *) NULL); |
172 | } | |
173 | ||
174 | const wxCharBuffer wxMBConv::cWC2MB(const wchar_t *psz) const | |
175 | { | |
f6bcfd97 | 176 | if (psz) |
6001e347 RR |
177 | { |
178 | size_t nLen = WC2MB((char *) NULL, psz, 0); | |
f6bcfd97 BP |
179 | if (nLen == (size_t)-1) |
180 | return wxCharBuffer((char *) NULL); | |
6001e347 RR |
181 | wxCharBuffer buf(nLen); |
182 | WC2MB((char *)(const char *) buf, psz, nLen); | |
183 | return buf; | |
f6bcfd97 BP |
184 | } |
185 | else | |
6001e347 RR |
186 | return wxCharBuffer((char *) NULL); |
187 | } | |
188 | ||
f6bcfd97 | 189 | // ---------------------------------------------------------------------------- |
6001e347 | 190 | // standard file conversion |
f6bcfd97 | 191 | // ---------------------------------------------------------------------------- |
6001e347 RR |
192 | |
193 | WXDLLEXPORT_DATA(wxMBConvFile) wxConvFile; | |
194 | ||
195 | // just use the libc conversion for now | |
196 | size_t wxMBConvFile::MB2WC(wchar_t *buf, const char *psz, size_t n) const | |
197 | { | |
198 | return wxMB2WC(buf, psz, n); | |
199 | } | |
200 | ||
201 | size_t wxMBConvFile::WC2MB(char *buf, const wchar_t *psz, size_t n) const | |
202 | { | |
203 | return wxWC2MB(buf, psz, n); | |
204 | } | |
205 | ||
f6bcfd97 | 206 | // ---------------------------------------------------------------------------- |
6001e347 | 207 | // standard gdk conversion |
f6bcfd97 BP |
208 | // ---------------------------------------------------------------------------- |
209 | ||
210 | #ifdef __WXGTK12__ | |
6001e347 RR |
211 | |
212 | WXDLLEXPORT_DATA(wxMBConvGdk) wxConvGdk; | |
213 | ||
214 | #include <gdk/gdk.h> | |
215 | ||
216 | size_t wxMBConvGdk::MB2WC(wchar_t *buf, const char *psz, size_t n) const | |
217 | { | |
dccce9ea | 218 | if (buf) |
4def3b35 VS |
219 | { |
220 | return gdk_mbstowcs((GdkWChar *)buf, psz, n); | |
dccce9ea VZ |
221 | } |
222 | else | |
4def3b35 VS |
223 | { |
224 | GdkWChar *nbuf = new GdkWChar[n=strlen(psz)]; | |
225 | size_t len = gdk_mbstowcs(nbuf, psz, n); | |
226 | delete[] nbuf; | |
227 | return len; | |
228 | } | |
6001e347 RR |
229 | } |
230 | ||
231 | size_t wxMBConvGdk::WC2MB(char *buf, const wchar_t *psz, size_t n) const | |
232 | { | |
4def3b35 VS |
233 | char *mbstr = gdk_wcstombs((GdkWChar *)psz); |
234 | size_t len = mbstr ? strlen(mbstr) : 0; | |
dccce9ea | 235 | if (buf) |
4def3b35 | 236 | { |
dccce9ea | 237 | if (len > n) |
4def3b35 VS |
238 | len = n; |
239 | memcpy(buf, psz, len); | |
dccce9ea | 240 | if (len < n) |
4def3b35 VS |
241 | buf[len] = 0; |
242 | } | |
243 | return len; | |
6001e347 | 244 | } |
f6bcfd97 | 245 | |
6001e347 RR |
246 | #endif // GTK > 1.0 |
247 | ||
248 | // ---------------------------------------------------------------------------- | |
249 | // UTF-7 | |
250 | // ---------------------------------------------------------------------------- | |
251 | ||
252 | WXDLLEXPORT_DATA(wxMBConvUTF7) wxConvUTF7; | |
253 | ||
254 | #if 0 | |
255 | static char utf7_setD[]="ABCDEFGHIJKLMNOPQRSTUVWXYZ" | |
256 | "abcdefghijklmnopqrstuvwxyz" | |
257 | "0123456789'(),-./:?"; | |
258 | static char utf7_setO[]="!\"#$%&*;<=>@[]^_`{|}"; | |
259 | static char utf7_setB[]="ABCDEFGHIJKLMNOPQRSTUVWXYZ" | |
260 | "abcdefghijklmnopqrstuvwxyz" | |
261 | "0123456789+/"; | |
262 | #endif | |
263 | ||
264 | // TODO: write actual implementations of UTF-7 here | |
265 | size_t wxMBConvUTF7::MB2WC(wchar_t * WXUNUSED(buf), | |
266 | const char * WXUNUSED(psz), | |
267 | size_t WXUNUSED(n)) const | |
268 | { | |
269 | return 0; | |
270 | } | |
271 | ||
272 | size_t wxMBConvUTF7::WC2MB(char * WXUNUSED(buf), | |
273 | const wchar_t * WXUNUSED(psz), | |
274 | size_t WXUNUSED(n)) const | |
275 | { | |
276 | return 0; | |
277 | } | |
278 | ||
f6bcfd97 | 279 | // ---------------------------------------------------------------------------- |
6001e347 | 280 | // UTF-8 |
f6bcfd97 | 281 | // ---------------------------------------------------------------------------- |
6001e347 RR |
282 | |
283 | WXDLLEXPORT_DATA(wxMBConvUTF8) wxConvUTF8; | |
284 | ||
dccce9ea | 285 | static wxUint32 utf8_max[]= |
4def3b35 | 286 | { 0x7f, 0x7ff, 0xffff, 0x1fffff, 0x3ffffff, 0x7fffffff, 0xffffffff }; |
6001e347 RR |
287 | |
288 | size_t wxMBConvUTF8::MB2WC(wchar_t *buf, const char *psz, size_t n) const | |
289 | { | |
4def3b35 VS |
290 | size_t len = 0; |
291 | ||
dccce9ea | 292 | while (*psz && ((!buf) || (len < n))) |
4def3b35 VS |
293 | { |
294 | unsigned char cc = *psz++, fc = cc; | |
295 | unsigned cnt; | |
dccce9ea | 296 | for (cnt = 0; fc & 0x80; cnt++) |
4def3b35 | 297 | fc <<= 1; |
dccce9ea | 298 | if (!cnt) |
4def3b35 VS |
299 | { |
300 | // plain ASCII char | |
dccce9ea | 301 | if (buf) |
4def3b35 VS |
302 | *buf++ = cc; |
303 | len++; | |
dccce9ea VZ |
304 | } |
305 | else | |
4def3b35 VS |
306 | { |
307 | cnt--; | |
dccce9ea | 308 | if (!cnt) |
4def3b35 VS |
309 | { |
310 | // invalid UTF-8 sequence | |
311 | return (size_t)-1; | |
dccce9ea VZ |
312 | } |
313 | else | |
4def3b35 VS |
314 | { |
315 | unsigned ocnt = cnt - 1; | |
316 | wxUint32 res = cc & (0x3f >> cnt); | |
dccce9ea | 317 | while (cnt--) |
4def3b35 VS |
318 | { |
319 | cc = *psz++; | |
dccce9ea | 320 | if ((cc & 0xC0) != 0x80) |
4def3b35 VS |
321 | { |
322 | // invalid UTF-8 sequence | |
323 | return (size_t)-1; | |
324 | } | |
325 | res = (res << 6) | (cc & 0x3f); | |
326 | } | |
dccce9ea | 327 | if (res <= utf8_max[ocnt]) |
4def3b35 VS |
328 | { |
329 | // illegal UTF-8 encoding | |
330 | return (size_t)-1; | |
331 | } | |
1cd52418 | 332 | #ifdef WC_UTF16 |
4def3b35 VS |
333 | size_t pa = encode_utf16(res, buf); |
334 | if (pa == (size_t)-1) | |
335 | return (size_t)-1; | |
dccce9ea | 336 | if (buf) |
4def3b35 VS |
337 | buf += pa; |
338 | len += pa; | |
1cd52418 | 339 | #else |
dccce9ea | 340 | if (buf) |
4def3b35 VS |
341 | *buf++ = res; |
342 | len++; | |
1cd52418 | 343 | #endif |
4def3b35 VS |
344 | } |
345 | } | |
6001e347 | 346 | } |
dccce9ea | 347 | if (buf && (len < n)) |
4def3b35 VS |
348 | *buf = 0; |
349 | return len; | |
6001e347 RR |
350 | } |
351 | ||
352 | size_t wxMBConvUTF8::WC2MB(char *buf, const wchar_t *psz, size_t n) const | |
353 | { | |
4def3b35 | 354 | size_t len = 0; |
6001e347 | 355 | |
dccce9ea | 356 | while (*psz && ((!buf) || (len < n))) |
4def3b35 VS |
357 | { |
358 | wxUint32 cc; | |
1cd52418 | 359 | #ifdef WC_UTF16 |
eccf1b2c | 360 | size_t pa = decode_utf16(psz, cc); |
4def3b35 | 361 | psz += (pa == (size_t)-1) ? 1 : pa; |
1cd52418 | 362 | #else |
4def3b35 VS |
363 | cc=(*psz++) & 0x7fffffff; |
364 | #endif | |
365 | unsigned cnt; | |
366 | for (cnt = 0; cc > utf8_max[cnt]; cnt++) {} | |
dccce9ea | 367 | if (!cnt) |
4def3b35 VS |
368 | { |
369 | // plain ASCII char | |
dccce9ea | 370 | if (buf) |
4def3b35 VS |
371 | *buf++ = cc; |
372 | len++; | |
dccce9ea VZ |
373 | } |
374 | ||
375 | else | |
4def3b35 VS |
376 | { |
377 | len += cnt + 1; | |
dccce9ea | 378 | if (buf) |
4def3b35 VS |
379 | { |
380 | *buf++ = (-128 >> cnt) | ((cc >> (cnt * 6)) & (0x3f >> cnt)); | |
381 | while (cnt--) | |
382 | *buf++ = 0x80 | ((cc >> (cnt * 6)) & 0x3f); | |
383 | } | |
384 | } | |
6001e347 | 385 | } |
4def3b35 VS |
386 | |
387 | if (buf && (len<n)) *buf = 0; | |
388 | return len; | |
6001e347 RR |
389 | } |
390 | ||
391 | // ---------------------------------------------------------------------------- | |
392 | // specified character set | |
393 | // ---------------------------------------------------------------------------- | |
394 | ||
f6bcfd97 BP |
395 | WXDLLEXPORT_DATA(wxCSConv) wxConvLocal((const wxChar *)NULL); |
396 | ||
397 | #include "wx/encconv.h" | |
398 | #include "wx/fontmap.h" | |
6001e347 | 399 | |
1cd52418 OK |
400 | // TODO: add some tables here |
401 | // - perhaps common encodings to common codepages (for Win32) | |
402 | // - perhaps common encodings to objects ("UTF8" -> wxConvUTF8) | |
403 | // - move wxEncodingConverter meat in here | |
404 | ||
04ef50df | 405 | #if defined(__WIN32__) && !defined(__WXMICROWIN__) |
1cd52418 OK |
406 | #include "wx/msw/registry.h" |
407 | // this should work if M$ Internet Exploiter is installed | |
408 | static long CharsetToCodepage(const wxChar *name) | |
409 | { | |
dccce9ea | 410 | if (!name) |
f1339c56 | 411 | return GetACP(); |
dccce9ea | 412 | |
f1339c56 | 413 | long CP=-1; |
dccce9ea | 414 | |
f1339c56 RR |
415 | wxString cn(name); |
416 | do { | |
5ce0e4ac | 417 | wxString path(wxT("MIME\\Database\\Charset\\")); |
f1339c56 | 418 | path += cn; |
5ce0e4ac VS |
419 | wxRegKey key(wxRegKey::HKCR, path); |
420 | ||
9c904e25 | 421 | if (!key.Exists()) break; |
dccce9ea | 422 | |
5ce0e4ac VS |
423 | // two cases: either there's an AliasForCharset string, |
424 | // or there are Codepage and InternetEncoding dwords. | |
425 | // The InternetEncoding gives us the actual encoding, | |
426 | // the Codepage just says which Windows character set to | |
427 | // use when displaying the data. | |
428 | if (key.HasValue(wxT("InternetEncoding")) && | |
429 | key.QueryValue(wxT("InternetEncoding"), &CP)) break; | |
dccce9ea | 430 | |
f1339c56 | 431 | // no encoding, see if it's an alias |
5ce0e4ac VS |
432 | if (!key.HasValue(wxT("AliasForCharset")) || |
433 | !key.QueryValue(wxT("AliasForCharset"), cn)) break; | |
f1339c56 | 434 | } while (1); |
dccce9ea | 435 | |
f1339c56 | 436 | return CP; |
1cd52418 OK |
437 | } |
438 | #endif | |
439 | ||
6001e347 RR |
440 | class wxCharacterSet |
441 | { | |
1cd52418 | 442 | public: |
f1339c56 RR |
443 | wxCharacterSet(const wxChar*name) |
444 | : cname(name) {} | |
445 | virtual ~wxCharacterSet() | |
446 | {} | |
dccce9ea | 447 | virtual size_t MB2WC(wchar_t *buf, const char *psz, size_t n) |
f1339c56 | 448 | { return (size_t)-1; } |
4def3b35 | 449 | virtual size_t WC2MB(char *buf, const wchar_t *psz, size_t n) |
f1339c56 RR |
450 | { return (size_t)-1; } |
451 | virtual bool usable() | |
452 | { return FALSE; } | |
453 | public: | |
454 | const wxChar*cname; | |
1cd52418 OK |
455 | }; |
456 | ||
457 | class ID_CharSet : public wxCharacterSet | |
458 | { | |
459 | public: | |
4def3b35 | 460 | ID_CharSet(const wxChar *name,wxMBConv *cnv) |
f1339c56 | 461 | : wxCharacterSet(name), work(cnv) {} |
dccce9ea | 462 | |
4def3b35 | 463 | size_t MB2WC(wchar_t *buf, const char *psz, size_t n) |
f1339c56 | 464 | { return work ? work->MB2WC(buf,psz,n) : (size_t)-1; } |
dccce9ea | 465 | |
4def3b35 | 466 | size_t WC2MB(char *buf, const wchar_t *psz, size_t n) |
f1339c56 | 467 | { return work ? work->WC2MB(buf,psz,n) : (size_t)-1; } |
dccce9ea | 468 | |
f1339c56 RR |
469 | bool usable() |
470 | { return work!=NULL; } | |
471 | public: | |
472 | wxMBConv*work; | |
1cd52418 OK |
473 | }; |
474 | ||
3caec1bb | 475 | |
1cd52418 | 476 | #ifdef HAVE_ICONV_H |
3caec1bb VS |
477 | |
478 | // VS: glibc 2.1.3 is broken in that iconv() conversion to/from UCS4 fails with E2BIG | |
479 | // if output buffer is _exactly_ as big as needed. Such case is (unless there's | |
480 | // yet another bug in glibc) the only case when iconv() returns with (size_t)-1 | |
481 | // (which means error) and says there are 0 bytes left in the input buffer -- | |
482 | // when _real_ error occurs, bytes-left-in-input buffer is non-zero. Hence, | |
483 | // this alternative test for iconv() failure. | |
484 | // [This bug does not appear in glibc 2.2.] | |
485 | #if defined(__GLIBC__) && __GLIBC__ == 2 && __GLIBC_MINOR__ <= 1 | |
486 | #define ICONV_FAILED(cres, bufLeft) ((cres == (size_t)-1) && \ | |
487 | (errno != E2BIG || bufLeft != 0)) | |
488 | #else | |
489 | #define ICONV_FAILED(cres, bufLeft) (cres == (size_t)-1) | |
490 | #endif | |
491 | ||
1cd52418 OK |
492 | class IC_CharSet : public wxCharacterSet |
493 | { | |
494 | public: | |
dccce9ea | 495 | IC_CharSet(const wxChar *name) |
3caec1bb | 496 | : wxCharacterSet(name) |
f1339c56 | 497 | { |
3caec1bb VS |
498 | m2w = iconv_open(WC_NAME, wxConvLibc.cWX2MB(cname)); |
499 | w2m = iconv_open(wxConvLibc.cWX2MB(cname), WC_NAME); | |
f1339c56 | 500 | } |
dccce9ea | 501 | |
3caec1bb VS |
502 | ~IC_CharSet() |
503 | { | |
dccce9ea | 504 | if ( m2w != (iconv_t)-1 ) |
3caec1bb | 505 | iconv_close(m2w); |
dccce9ea | 506 | if ( w2m != (iconv_t)-1 ) |
3caec1bb | 507 | iconv_close(w2m); |
f1339c56 | 508 | } |
dccce9ea | 509 | |
3caec1bb | 510 | size_t MB2WC(wchar_t *buf, const char *psz, size_t n) |
f1339c56 | 511 | { |
f1339c56 | 512 | size_t inbuf = strlen(psz); |
3caec1bb | 513 | size_t outbuf = n * SIZEOF_WCHAR_T; |
f1339c56 | 514 | size_t res, cres; |
3caec1bb VS |
515 | // VS: Use these instead of psz, buf because iconv() modifies its arguments: |
516 | wchar_t *bufPtr = buf; | |
517 | const char *pszPtr = psz; | |
518 | ||
f1339c56 RR |
519 | if (buf) |
520 | { | |
521 | // have destination buffer, convert there | |
95c8801c | 522 | #ifdef WX_ICONV_TAKES_CHAR |
3caec1bb | 523 | cres = iconv(m2w, (char**)&pszPtr, &inbuf, (char**)&bufPtr, &outbuf); |
95c8801c | 524 | #else |
3caec1bb | 525 | cres = iconv(m2w, &pszPtr, &inbuf, (char**)&bufPtr, &outbuf); |
95c8801c | 526 | #endif |
3caec1bb | 527 | res = n - (outbuf / SIZEOF_WCHAR_T); |
f1339c56 | 528 | // convert to native endianness |
3caec1bb VS |
529 | #ifdef WC_NEED_BSWAP |
530 | WC_BSWAP(buf /* _not_ bufPtr */, res) | |
531 | #endif | |
f1339c56 RR |
532 | } |
533 | else | |
534 | { | |
535 | // no destination buffer... convert using temp buffer | |
536 | // to calculate destination buffer requirement | |
537 | wchar_t tbuf[8]; | |
538 | res = 0; | |
539 | do { | |
3caec1bb | 540 | bufPtr = tbuf; outbuf = 8*SIZEOF_WCHAR_T; |
95c8801c | 541 | #ifdef WX_ICONV_TAKES_CHAR |
3caec1bb | 542 | cres = iconv( m2w, (char**)&pszPtr, &inbuf, (char**)&bufPtr, &outbuf ); |
95c8801c | 543 | #else |
3caec1bb | 544 | cres = iconv( m2w, &pszPtr, &inbuf, (char**)&bufPtr, &outbuf ); |
95c8801c | 545 | #endif |
f1339c56 RR |
546 | res += 8-(outbuf/SIZEOF_WCHAR_T); |
547 | } while ((cres==(size_t)-1) && (errno==E2BIG)); | |
548 | } | |
dccce9ea | 549 | |
3caec1bb | 550 | if (ICONV_FAILED(cres, inbuf)) |
f1339c56 | 551 | return (size_t)-1; |
3caec1bb | 552 | |
f1339c56 RR |
553 | return res; |
554 | } | |
dccce9ea | 555 | |
4def3b35 | 556 | size_t WC2MB(char *buf, const wchar_t *psz, size_t n) |
f1339c56 | 557 | { |
1cd52418 | 558 | #if defined(__BORLANDC__) && (__BORLANDC__ > 0x530) |
3caec1bb | 559 | size_t inbuf = std::wcslen(psz) * SIZEOF_WCHAR_T; |
1cd52418 | 560 | #else |
3caec1bb | 561 | size_t inbuf = ::wcslen(psz) * SIZEOF_WCHAR_T; |
1cd52418 | 562 | #endif |
f1339c56 RR |
563 | size_t outbuf = n; |
564 | size_t res, cres; | |
3caec1bb | 565 | |
1cd52418 | 566 | #ifdef WC_NEED_BSWAP |
f1339c56 RR |
567 | // need to copy to temp buffer to switch endianness |
568 | // this absolutely doesn't rock! | |
569 | // (no, doing WC_BSWAP twice on the original buffer won't help, as it | |
570 | // could be in read-only memory, or be accessed in some other thread) | |
4def3b35 | 571 | wchar_t *tmpbuf=(wchar_t*)malloc((inbuf+1)*SIZEOF_WCHAR_T); |
f1339c56 RR |
572 | memcpy(tmpbuf,psz,(inbuf+1)*SIZEOF_WCHAR_T); |
573 | WC_BSWAP(tmpbuf, inbuf) | |
574 | psz=tmpbuf; | |
1cd52418 | 575 | #endif |
f1339c56 RR |
576 | if (buf) |
577 | { | |
578 | // have destination buffer, convert there | |
95c8801c | 579 | #ifdef WX_ICONV_TAKES_CHAR |
f1339c56 | 580 | cres = iconv( w2m, (char**)&psz, &inbuf, &buf, &outbuf ); |
95c8801c VS |
581 | #else |
582 | cres = iconv( w2m, (const char**)&psz, &inbuf, &buf, &outbuf ); | |
583 | #endif | |
f1339c56 RR |
584 | res = n-outbuf; |
585 | } | |
586 | else | |
587 | { | |
588 | // no destination buffer... convert using temp buffer | |
589 | // to calculate destination buffer requirement | |
590 | char tbuf[16]; | |
591 | res = 0; | |
592 | do { | |
593 | buf = tbuf; outbuf = 16; | |
95c8801c | 594 | #ifdef WX_ICONV_TAKES_CHAR |
f1339c56 | 595 | cres = iconv( w2m, (char**)&psz, &inbuf, &buf, &outbuf ); |
95c8801c VS |
596 | #else |
597 | cres = iconv( w2m, (const char**)&psz, &inbuf, &buf, &outbuf ); | |
598 | #endif | |
f1339c56 RR |
599 | res += 16 - outbuf; |
600 | } while ((cres==(size_t)-1) && (errno==E2BIG)); | |
601 | } | |
1cd52418 | 602 | #ifdef WC_NEED_BSWAP |
f1339c56 | 603 | free(tmpbuf); |
1cd52418 | 604 | #endif |
3caec1bb | 605 | if (ICONV_FAILED(cres, inbuf)) |
f1339c56 | 606 | return (size_t)-1; |
dccce9ea | 607 | |
f1339c56 RR |
608 | return res; |
609 | } | |
dccce9ea | 610 | |
f1339c56 | 611 | bool usable() |
3caec1bb | 612 | { return (m2w != (iconv_t)-1) && (w2m != (iconv_t)-1); } |
dccce9ea | 613 | |
f1339c56 RR |
614 | public: |
615 | iconv_t m2w, w2m; | |
1cd52418 OK |
616 | }; |
617 | #endif | |
618 | ||
04ef50df | 619 | #if defined(__WIN32__) && !defined(__WXMICROWIN__) |
1cd52418 OK |
620 | class CP_CharSet : public wxCharacterSet |
621 | { | |
622 | public: | |
dccce9ea | 623 | CP_CharSet(const wxChar*name) |
f1339c56 | 624 | : wxCharacterSet(name), CodePage(CharsetToCodepage(name)) {} |
dccce9ea | 625 | |
4def3b35 | 626 | size_t MB2WC(wchar_t *buf, const char *psz, size_t n) |
f1339c56 | 627 | { |
dccce9ea | 628 | size_t len = |
4def3b35 | 629 | MultiByteToWideChar(CodePage, 0, psz, -1, buf, buf ? n : 0); |
1e6feb95 | 630 | //VS: returns # of written chars for buf!=NULL and *size* |
35d764b0 VS |
631 | // needed buffer for buf==NULL |
632 | return len ? (buf ? len : len-1) : (size_t)-1; | |
f1339c56 | 633 | } |
dccce9ea | 634 | |
4def3b35 | 635 | size_t WC2MB(char *buf, const wchar_t *psz, size_t n) |
f1339c56 | 636 | { |
4def3b35 VS |
637 | size_t len = WideCharToMultiByte(CodePage, 0, psz, -1, buf, |
638 | buf ? n : 0, NULL, NULL); | |
1e6feb95 | 639 | //VS: returns # of written chars for buf!=NULL and *size* |
35d764b0 VS |
640 | // needed buffer for buf==NULL |
641 | return len ? (buf ? len : len-1) : (size_t)-1; | |
f1339c56 | 642 | } |
dccce9ea | 643 | |
f1339c56 | 644 | bool usable() |
4def3b35 | 645 | { return CodePage != -1; } |
f1339c56 RR |
646 | |
647 | public: | |
648 | long CodePage; | |
1cd52418 | 649 | }; |
1e6feb95 VZ |
650 | #endif // __WIN32__ |
651 | ||
652 | #if wxUSE_FONTMAP | |
1cd52418 OK |
653 | |
654 | class EC_CharSet : public wxCharacterSet | |
655 | { | |
6001e347 | 656 | public: |
f1339c56 RR |
657 | // temporarily just use wxEncodingConverter stuff, |
658 | // so that it works while a better implementation is built | |
dccce9ea | 659 | EC_CharSet(const wxChar*name) : wxCharacterSet(name), |
4def3b35 | 660 | enc(wxFONTENCODING_SYSTEM) |
f1339c56 RR |
661 | { |
662 | if (name) | |
663 | enc = wxTheFontMapper->CharsetToEncoding(name, FALSE); | |
664 | m2w.Init(enc, wxFONTENCODING_UNICODE); | |
665 | w2m.Init(wxFONTENCODING_UNICODE, enc); | |
666 | } | |
dccce9ea | 667 | |
4def3b35 | 668 | size_t MB2WC(wchar_t *buf, const char *psz, size_t n) |
f1339c56 RR |
669 | { |
670 | size_t inbuf = strlen(psz); | |
dccce9ea | 671 | if (buf) |
4def3b35 | 672 | m2w.Convert(psz,buf); |
f1339c56 RR |
673 | return inbuf; |
674 | } | |
dccce9ea | 675 | |
4def3b35 | 676 | size_t WC2MB(char *buf, const wchar_t *psz, size_t n) |
f1339c56 | 677 | { |
1cd52418 | 678 | #if defined(__BORLANDC__) && (__BORLANDC__ > 0x530) |
f1339c56 | 679 | size_t inbuf = std::wcslen(psz); |
1cd52418 | 680 | #else |
f1339c56 | 681 | size_t inbuf = ::wcslen(psz); |
1cd52418 | 682 | #endif |
f1339c56 RR |
683 | if (buf) |
684 | w2m.Convert(psz,buf); | |
dccce9ea | 685 | |
f1339c56 RR |
686 | return inbuf; |
687 | } | |
dccce9ea | 688 | |
f1339c56 RR |
689 | bool usable() |
690 | { return (enc!=wxFONTENCODING_SYSTEM) && (enc!=wxFONTENCODING_DEFAULT); } | |
691 | ||
692 | public: | |
693 | wxFontEncoding enc; | |
694 | wxEncodingConverter m2w, w2m; | |
f6bcfd97 | 695 | }; |
6001e347 | 696 | |
1e6feb95 VZ |
697 | #endif // wxUSE_FONTMAP |
698 | ||
f6bcfd97 | 699 | static wxCharacterSet *wxGetCharacterSet(const wxChar *name) |
6001e347 | 700 | { |
f1339c56 RR |
701 | wxCharacterSet *cset = NULL; |
702 | if (name) | |
703 | { | |
4def3b35 | 704 | if (wxStricmp(name, wxT("UTF8")) == 0 || wxStricmp(name, wxT("UTF-8")) == 0) |
f1339c56 RR |
705 | { |
706 | cset = new ID_CharSet(name, &wxConvUTF8); | |
707 | } | |
708 | else | |
709 | { | |
1cd52418 | 710 | #ifdef HAVE_ICONV_H |
f1339c56 | 711 | cset = new IC_CharSet(name); // may not take NULL |
1cd52418 | 712 | #endif |
f1339c56 | 713 | } |
1cd52418 | 714 | } |
dccce9ea | 715 | |
1e6feb95 VZ |
716 | if (cset && cset->usable()) |
717 | return cset; | |
718 | ||
dccce9ea VZ |
719 | if (cset) |
720 | { | |
721 | delete cset; | |
722 | cset = NULL; | |
723 | } | |
724 | ||
04ef50df | 725 | #if defined(__WIN32__) && !defined(__WXMICROWIN__) |
f1339c56 | 726 | cset = new CP_CharSet(name); // may take NULL |
dccce9ea VZ |
727 | if (cset->usable()) |
728 | return cset; | |
729 | ||
730 | delete cset; | |
731 | #endif // __WIN32__ | |
732 | ||
1e6feb95 | 733 | #if wxUSE_FONTMAP |
f1339c56 | 734 | cset = new EC_CharSet(name); |
dccce9ea VZ |
735 | if (cset->usable()) |
736 | return cset; | |
1e6feb95 | 737 | #endif // wxUSE_FONTMAP |
dccce9ea | 738 | |
f1339c56 | 739 | delete cset; |
3caec1bb | 740 | wxLogError(_("Unknown encoding '%s'!"), name); |
f1339c56 | 741 | return NULL; |
6001e347 RR |
742 | } |
743 | ||
6001e347 RR |
744 | wxCSConv::wxCSConv(const wxChar *charset) |
745 | { | |
dccce9ea | 746 | m_name = (wxChar *)NULL; |
f1339c56 | 747 | m_cset = (wxCharacterSet *) NULL; |
82713003 VZ |
748 | m_deferred = TRUE; |
749 | ||
f1339c56 | 750 | SetName(charset); |
6001e347 RR |
751 | } |
752 | ||
753 | wxCSConv::~wxCSConv() | |
754 | { | |
dccce9ea VZ |
755 | free(m_name); |
756 | delete m_cset; | |
6001e347 RR |
757 | } |
758 | ||
759 | void wxCSConv::SetName(const wxChar *charset) | |
760 | { | |
f1339c56 RR |
761 | if (charset) |
762 | { | |
763 | m_name = wxStrdup(charset); | |
764 | m_deferred = TRUE; | |
765 | } | |
6001e347 RR |
766 | } |
767 | ||
768 | void wxCSConv::LoadNow() | |
769 | { | |
f1339c56 RR |
770 | if (m_deferred) |
771 | { | |
dccce9ea | 772 | if ( !m_name ) |
f1339c56 | 773 | { |
dccce9ea VZ |
774 | wxString name = wxLocale::GetSystemEncodingName(); |
775 | if ( !name.empty() ) | |
776 | SetName(name); | |
f1339c56 | 777 | } |
dccce9ea | 778 | |
a45a98fb VZ |
779 | // wxGetCharacterSet() complains about NULL name |
780 | m_cset = m_name ? wxGetCharacterSet(m_name) : NULL; | |
f1339c56 | 781 | m_deferred = FALSE; |
6001e347 | 782 | } |
6001e347 RR |
783 | } |
784 | ||
785 | size_t wxCSConv::MB2WC(wchar_t *buf, const char *psz, size_t n) const | |
786 | { | |
f1339c56 | 787 | ((wxCSConv *)this)->LoadNow(); // discard constness |
dccce9ea | 788 | |
f1339c56 RR |
789 | if (m_cset) |
790 | return m_cset->MB2WC(buf, psz, n); | |
791 | ||
792 | // latin-1 (direct) | |
4def3b35 | 793 | size_t len = strlen(psz); |
dccce9ea | 794 | |
f1339c56 RR |
795 | if (buf) |
796 | { | |
4def3b35 | 797 | for (size_t c = 0; c <= len; c++) |
f1339c56 RR |
798 | buf[c] = (unsigned char)(psz[c]); |
799 | } | |
dccce9ea | 800 | |
f1339c56 | 801 | return len; |
6001e347 RR |
802 | } |
803 | ||
804 | size_t wxCSConv::WC2MB(char *buf, const wchar_t *psz, size_t n) const | |
805 | { | |
f1339c56 | 806 | ((wxCSConv *)this)->LoadNow(); // discard constness |
dccce9ea | 807 | |
f1339c56 RR |
808 | if (m_cset) |
809 | return m_cset->WC2MB(buf, psz, n); | |
1cd52418 | 810 | |
f1339c56 | 811 | // latin-1 (direct) |
d834f22c | 812 | #if defined(__BORLANDC__) && (__BORLANDC__ > 0x530) |
f1339c56 | 813 | size_t len=std::wcslen(psz); |
d834f22c | 814 | #else |
f1339c56 | 815 | size_t len=::wcslen(psz); |
d834f22c | 816 | #endif |
f1339c56 RR |
817 | if (buf) |
818 | { | |
4def3b35 VS |
819 | for (size_t c = 0; c <= len; c++) |
820 | buf[c] = (psz[c] > 0xff) ? '?' : psz[c]; | |
f1339c56 | 821 | } |
dccce9ea | 822 | |
f1339c56 | 823 | return len; |
6001e347 RR |
824 | } |
825 | ||
1cd52418 | 826 | #ifdef HAVE_ICONV_H |
dccce9ea | 827 | |
1cd52418 OK |
828 | class IC_CharSetConverter |
829 | { | |
830 | public: | |
4def3b35 VS |
831 | IC_CharSetConverter(IC_CharSet *from, IC_CharSet *to) |
832 | { | |
dccce9ea VZ |
833 | cnv = iconv_open(wxConvLibc.cWX2MB(to->cname), |
834 | wxConvLibc.cWX2MB(from->cname)); | |
4def3b35 | 835 | } |
dccce9ea | 836 | |
f1339c56 | 837 | ~IC_CharSetConverter() |
dccce9ea VZ |
838 | { |
839 | if (cnv != (iconv_t)-1) | |
840 | iconv_close(cnv); | |
4def3b35 | 841 | } |
dccce9ea | 842 | |
4def3b35 | 843 | size_t Convert(char *buf, const char *psz, size_t n) |
f1339c56 RR |
844 | { |
845 | size_t inbuf = strlen(psz); | |
846 | size_t outbuf = n; | |
95c8801c | 847 | #ifdef WX_ICONV_TAKES_CHAR |
f1339c56 | 848 | size_t res = iconv( cnv, (char**)&psz, &inbuf, &buf, &outbuf ); |
95c8801c VS |
849 | #else |
850 | size_t res = iconv( cnv, &psz, &inbuf, &buf, &outbuf ); | |
851 | #endif | |
dccce9ea | 852 | if (res == (size_t)-1) |
4def3b35 VS |
853 | return (size_t)-1; |
854 | return (n - outbuf); | |
f1339c56 RR |
855 | } |
856 | ||
857 | public: | |
858 | iconv_t cnv; | |
1cd52418 | 859 | }; |
dccce9ea VZ |
860 | |
861 | #endif // HAVE_ICONV_H | |
1cd52418 OK |
862 | |
863 | class EC_CharSetConverter | |
864 | { | |
865 | public: | |
1e6feb95 | 866 | EC_CharSetConverter(EC_CharSet* from,EC_CharSet* to) |
f1339c56 | 867 | { cnv.Init(from->enc,to->enc); } |
dccce9ea | 868 | |
1e6feb95 | 869 | size_t Convert(char* buf, const char* psz, size_t n) |
f1339c56 RR |
870 | { |
871 | size_t inbuf = strlen(psz); | |
872 | if (buf) cnv.Convert(psz,buf); | |
873 | return inbuf; | |
874 | } | |
dccce9ea | 875 | |
f1339c56 RR |
876 | public: |
877 | wxEncodingConverter cnv; | |
1cd52418 OK |
878 | }; |
879 | ||
f6bcfd97 BP |
880 | #else // !wxUSE_WCHAR_T |
881 | ||
882 | // ---------------------------------------------------------------------------- | |
883 | // stand-ins in absence of wchar_t | |
884 | // ---------------------------------------------------------------------------- | |
885 | ||
886 | WXDLLEXPORT_DATA(wxMBConv) wxConvLibc, wxConvFile; | |
887 | ||
888 | #endif // wxUSE_WCHAR_T | |
6001e347 RR |
889 | |
890 |