| 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 | #else |
| 56 | // stand-ins in absence of wchar_t |
| 57 | WXDLLEXPORT_DATA(wxMBConv) wxConvLibc, wxConvFile; |
| 58 | #endif // wxUSE_WCHAR_T |
| 59 | |
| 60 | WXDLLEXPORT_DATA(wxMBConv *) wxConvCurrent = &wxConvLibc; |
| 61 | |
| 62 | class wxStrConvModule: public wxModule |
| 63 | { |
| 64 | public: |
| 65 | wxStrConvModule() : wxModule() { } |
| 66 | virtual bool OnInit() { return TRUE; } |
| 67 | virtual void OnExit() |
| 68 | { |
| 69 | #if wxUSE_WCHAR_T |
| 70 | wxConvLocal.Clear(); |
| 71 | #endif |
| 72 | } |
| 73 | |
| 74 | DECLARE_DYNAMIC_CLASS(wxStrConvModule) |
| 75 | }; |
| 76 | |
| 77 | IMPLEMENT_DYNAMIC_CLASS(wxStrConvModule, wxModule) |
| 78 | |
| 79 | |
| 80 | // ---------------------------------------------------------------------------- |
| 81 | // headers |
| 82 | // ---------------------------------------------------------------------------- |
| 83 | |
| 84 | #if wxUSE_WCHAR_T |
| 85 | |
| 86 | #ifdef __SALFORDC__ |
| 87 | #include <clib.h> |
| 88 | #endif |
| 89 | |
| 90 | #ifdef HAVE_ICONV |
| 91 | #include <iconv.h> |
| 92 | #endif |
| 93 | |
| 94 | #include "wx/encconv.h" |
| 95 | #include "wx/fontmap.h" |
| 96 | |
| 97 | // ---------------------------------------------------------------------------- |
| 98 | // macros |
| 99 | // ---------------------------------------------------------------------------- |
| 100 | |
| 101 | #define BSWAP_UCS4(str, len) { unsigned _c; for (_c=0; _c<len; _c++) str[_c]=wxUINT32_SWAP_ALWAYS(str[_c]); } |
| 102 | #define BSWAP_UTF16(str, len) { unsigned _c; for (_c=0; _c<len; _c++) str[_c]=wxUINT16_SWAP_ALWAYS(str[_c]); } |
| 103 | |
| 104 | // under Unix SIZEOF_WCHAR_T is defined by configure, but under other platforms |
| 105 | // it might be not defined - assume the most common value |
| 106 | #ifndef SIZEOF_WCHAR_T |
| 107 | #define SIZEOF_WCHAR_T 2 |
| 108 | #endif // !defined(SIZEOF_WCHAR_T) |
| 109 | |
| 110 | #if SIZEOF_WCHAR_T == 4 |
| 111 | #define WC_NAME "UCS4" |
| 112 | #define WC_BSWAP BSWAP_UCS4 |
| 113 | #ifdef WORDS_BIGENDIAN |
| 114 | #define WC_NAME_BEST "UCS-4BE" |
| 115 | #else |
| 116 | #define WC_NAME_BEST "UCS-4LE" |
| 117 | #endif |
| 118 | #elif SIZEOF_WCHAR_T == 2 |
| 119 | #define WC_NAME "UTF16" |
| 120 | #define WC_BSWAP BSWAP_UTF16 |
| 121 | #define WC_UTF16 |
| 122 | #ifdef WORDS_BIGENDIAN |
| 123 | #define WC_NAME_BEST "UTF-16BE" |
| 124 | #else |
| 125 | #define WC_NAME_BEST "UTF-16LE" |
| 126 | #endif |
| 127 | #else // sizeof(wchar_t) != 2 nor 4 |
| 128 | // I don't know what to do about this |
| 129 | #error "Weird sizeof(wchar_t): please report your platform details to wx-users mailing list" |
| 130 | #endif |
| 131 | |
| 132 | // ============================================================================ |
| 133 | // implementation |
| 134 | // ============================================================================ |
| 135 | |
| 136 | // ---------------------------------------------------------------------------- |
| 137 | // UTF-16 en/decoding |
| 138 | // ---------------------------------------------------------------------------- |
| 139 | |
| 140 | #ifdef WC_UTF16 |
| 141 | |
| 142 | static size_t encode_utf16(wxUint32 input, wchar_t *output) |
| 143 | { |
| 144 | if (input<=0xffff) |
| 145 | { |
| 146 | if (output) *output++ = (wchar_t) input; |
| 147 | return 1; |
| 148 | } |
| 149 | else if (input>=0x110000) |
| 150 | { |
| 151 | return (size_t)-1; |
| 152 | } |
| 153 | else |
| 154 | { |
| 155 | if (output) |
| 156 | { |
| 157 | *output++ = (wchar_t) ((input >> 10)+0xd7c0); |
| 158 | *output++ = (wchar_t) ((input&0x3ff)+0xdc00); |
| 159 | } |
| 160 | return 2; |
| 161 | } |
| 162 | } |
| 163 | |
| 164 | static size_t decode_utf16(const wchar_t* input, wxUint32& output) |
| 165 | { |
| 166 | if ((*input<0xd800) || (*input>0xdfff)) |
| 167 | { |
| 168 | output = *input; |
| 169 | return 1; |
| 170 | } |
| 171 | else if ((input[1]<0xdc00) || (input[1]>=0xdfff)) |
| 172 | { |
| 173 | output = *input; |
| 174 | return (size_t)-1; |
| 175 | } |
| 176 | else |
| 177 | { |
| 178 | output = ((input[0] - 0xd7c0) << 10) + (input[1] - 0xdc00); |
| 179 | return 2; |
| 180 | } |
| 181 | } |
| 182 | |
| 183 | #endif // WC_UTF16 |
| 184 | |
| 185 | // ---------------------------------------------------------------------------- |
| 186 | // wxMBConv |
| 187 | // ---------------------------------------------------------------------------- |
| 188 | |
| 189 | size_t wxMBConv::MB2WC(wchar_t *buf, const char *psz, size_t n) const |
| 190 | { |
| 191 | return wxMB2WC(buf, psz, n); |
| 192 | } |
| 193 | |
| 194 | size_t wxMBConv::WC2MB(char *buf, const wchar_t *psz, size_t n) const |
| 195 | { |
| 196 | return wxWC2MB(buf, psz, n); |
| 197 | } |
| 198 | |
| 199 | const wxWCharBuffer wxMBConv::cMB2WC(const char *psz) const |
| 200 | { |
| 201 | if (psz) |
| 202 | { |
| 203 | size_t nLen = MB2WC((wchar_t *) NULL, psz, 0); |
| 204 | if (nLen == (size_t)-1) |
| 205 | return wxWCharBuffer((wchar_t *) NULL); |
| 206 | wxWCharBuffer buf(nLen); |
| 207 | MB2WC((wchar_t *)(const wchar_t *) buf, psz, nLen); |
| 208 | return buf; |
| 209 | } |
| 210 | else |
| 211 | return wxWCharBuffer((wchar_t *) NULL); |
| 212 | } |
| 213 | |
| 214 | const wxCharBuffer wxMBConv::cWC2MB(const wchar_t *psz) const |
| 215 | { |
| 216 | if (psz) |
| 217 | { |
| 218 | size_t nLen = WC2MB((char *) NULL, psz, 0); |
| 219 | if (nLen == (size_t)-1) |
| 220 | return wxCharBuffer((char *) NULL); |
| 221 | wxCharBuffer buf(nLen); |
| 222 | WC2MB((char *)(const char *) buf, psz, nLen); |
| 223 | return buf; |
| 224 | } |
| 225 | else |
| 226 | return wxCharBuffer((char *) NULL); |
| 227 | } |
| 228 | |
| 229 | // ---------------------------------------------------------------------------- |
| 230 | // standard file conversion |
| 231 | // ---------------------------------------------------------------------------- |
| 232 | |
| 233 | WXDLLEXPORT_DATA(wxMBConvFile) wxConvFile; |
| 234 | |
| 235 | // just use the libc conversion for now |
| 236 | size_t wxMBConvFile::MB2WC(wchar_t *buf, const char *psz, size_t n) const |
| 237 | { |
| 238 | return wxMB2WC(buf, psz, n); |
| 239 | } |
| 240 | |
| 241 | size_t wxMBConvFile::WC2MB(char *buf, const wchar_t *psz, size_t n) const |
| 242 | { |
| 243 | return wxWC2MB(buf, psz, n); |
| 244 | } |
| 245 | |
| 246 | // ---------------------------------------------------------------------------- |
| 247 | // standard gdk conversion |
| 248 | // ---------------------------------------------------------------------------- |
| 249 | |
| 250 | #ifdef __WXGTK12__ |
| 251 | |
| 252 | WXDLLEXPORT_DATA(wxMBConvGdk) wxConvGdk; |
| 253 | |
| 254 | #include <gdk/gdk.h> |
| 255 | |
| 256 | size_t wxMBConvGdk::MB2WC(wchar_t *buf, const char *psz, size_t n) const |
| 257 | { |
| 258 | if (buf) |
| 259 | { |
| 260 | return gdk_mbstowcs((GdkWChar *)buf, psz, n); |
| 261 | } |
| 262 | else |
| 263 | { |
| 264 | GdkWChar *nbuf = new GdkWChar[n=strlen(psz)]; |
| 265 | size_t len = gdk_mbstowcs(nbuf, psz, n); |
| 266 | delete[] nbuf; |
| 267 | return len; |
| 268 | } |
| 269 | } |
| 270 | |
| 271 | size_t wxMBConvGdk::WC2MB(char *buf, const wchar_t *psz, size_t n) const |
| 272 | { |
| 273 | char *mbstr = gdk_wcstombs((GdkWChar *)psz); |
| 274 | size_t len = mbstr ? strlen(mbstr) : 0; |
| 275 | if (buf) |
| 276 | { |
| 277 | if (len > n) |
| 278 | len = n; |
| 279 | memcpy(buf, psz, len); |
| 280 | if (len < n) |
| 281 | buf[len] = 0; |
| 282 | } |
| 283 | return len; |
| 284 | } |
| 285 | |
| 286 | #endif // GTK > 1.0 |
| 287 | |
| 288 | // ---------------------------------------------------------------------------- |
| 289 | // UTF-7 |
| 290 | // ---------------------------------------------------------------------------- |
| 291 | |
| 292 | WXDLLEXPORT_DATA(wxMBConvUTF7) wxConvUTF7; |
| 293 | |
| 294 | #if 0 |
| 295 | static char utf7_setD[]="ABCDEFGHIJKLMNOPQRSTUVWXYZ" |
| 296 | "abcdefghijklmnopqrstuvwxyz" |
| 297 | "0123456789'(),-./:?"; |
| 298 | static char utf7_setO[]="!\"#$%&*;<=>@[]^_`{|}"; |
| 299 | static char utf7_setB[]="ABCDEFGHIJKLMNOPQRSTUVWXYZ" |
| 300 | "abcdefghijklmnopqrstuvwxyz" |
| 301 | "0123456789+/"; |
| 302 | #endif |
| 303 | |
| 304 | // TODO: write actual implementations of UTF-7 here |
| 305 | size_t wxMBConvUTF7::MB2WC(wchar_t * WXUNUSED(buf), |
| 306 | const char * WXUNUSED(psz), |
| 307 | size_t WXUNUSED(n)) const |
| 308 | { |
| 309 | return 0; |
| 310 | } |
| 311 | |
| 312 | size_t wxMBConvUTF7::WC2MB(char * WXUNUSED(buf), |
| 313 | const wchar_t * WXUNUSED(psz), |
| 314 | size_t WXUNUSED(n)) const |
| 315 | { |
| 316 | return 0; |
| 317 | } |
| 318 | |
| 319 | // ---------------------------------------------------------------------------- |
| 320 | // UTF-8 |
| 321 | // ---------------------------------------------------------------------------- |
| 322 | |
| 323 | WXDLLEXPORT_DATA(wxMBConvUTF8) wxConvUTF8; |
| 324 | |
| 325 | static wxUint32 utf8_max[]= |
| 326 | { 0x7f, 0x7ff, 0xffff, 0x1fffff, 0x3ffffff, 0x7fffffff, 0xffffffff }; |
| 327 | |
| 328 | size_t wxMBConvUTF8::MB2WC(wchar_t *buf, const char *psz, size_t n) const |
| 329 | { |
| 330 | size_t len = 0; |
| 331 | |
| 332 | while (*psz && ((!buf) || (len < n))) |
| 333 | { |
| 334 | unsigned char cc = *psz++, fc = cc; |
| 335 | unsigned cnt; |
| 336 | for (cnt = 0; fc & 0x80; cnt++) |
| 337 | fc <<= 1; |
| 338 | if (!cnt) |
| 339 | { |
| 340 | // plain ASCII char |
| 341 | if (buf) |
| 342 | *buf++ = cc; |
| 343 | len++; |
| 344 | } |
| 345 | else |
| 346 | { |
| 347 | cnt--; |
| 348 | if (!cnt) |
| 349 | { |
| 350 | // invalid UTF-8 sequence |
| 351 | return (size_t)-1; |
| 352 | } |
| 353 | else |
| 354 | { |
| 355 | unsigned ocnt = cnt - 1; |
| 356 | wxUint32 res = cc & (0x3f >> cnt); |
| 357 | while (cnt--) |
| 358 | { |
| 359 | cc = *psz++; |
| 360 | if ((cc & 0xC0) != 0x80) |
| 361 | { |
| 362 | // invalid UTF-8 sequence |
| 363 | return (size_t)-1; |
| 364 | } |
| 365 | res = (res << 6) | (cc & 0x3f); |
| 366 | } |
| 367 | if (res <= utf8_max[ocnt]) |
| 368 | { |
| 369 | // illegal UTF-8 encoding |
| 370 | return (size_t)-1; |
| 371 | } |
| 372 | #ifdef WC_UTF16 |
| 373 | size_t pa = encode_utf16(res, buf); |
| 374 | if (pa == (size_t)-1) |
| 375 | return (size_t)-1; |
| 376 | if (buf) |
| 377 | buf += pa; |
| 378 | len += pa; |
| 379 | #else // !WC_UTF16 |
| 380 | if (buf) |
| 381 | *buf++ = res; |
| 382 | len++; |
| 383 | #endif // WC_UTF16/!WC_UTF16 |
| 384 | } |
| 385 | } |
| 386 | } |
| 387 | if (buf && (len < n)) |
| 388 | *buf = 0; |
| 389 | return len; |
| 390 | } |
| 391 | |
| 392 | size_t wxMBConvUTF8::WC2MB(char *buf, const wchar_t *psz, size_t n) const |
| 393 | { |
| 394 | size_t len = 0; |
| 395 | |
| 396 | while (*psz && ((!buf) || (len < n))) |
| 397 | { |
| 398 | wxUint32 cc; |
| 399 | #ifdef WC_UTF16 |
| 400 | size_t pa = decode_utf16(psz, cc); |
| 401 | psz += (pa == (size_t)-1) ? 1 : pa; |
| 402 | #else |
| 403 | cc=(*psz++) & 0x7fffffff; |
| 404 | #endif |
| 405 | unsigned cnt; |
| 406 | for (cnt = 0; cc > utf8_max[cnt]; cnt++) {} |
| 407 | if (!cnt) |
| 408 | { |
| 409 | // plain ASCII char |
| 410 | if (buf) |
| 411 | *buf++ = (char) cc; |
| 412 | len++; |
| 413 | } |
| 414 | |
| 415 | else |
| 416 | { |
| 417 | len += cnt + 1; |
| 418 | if (buf) |
| 419 | { |
| 420 | *buf++ = (char) ((-128 >> cnt) | ((cc >> (cnt * 6)) & (0x3f >> cnt))); |
| 421 | while (cnt--) |
| 422 | *buf++ = (char) (0x80 | ((cc >> (cnt * 6)) & 0x3f)); |
| 423 | } |
| 424 | } |
| 425 | } |
| 426 | |
| 427 | if (buf && (len<n)) *buf = 0; |
| 428 | return len; |
| 429 | } |
| 430 | |
| 431 | // ============================================================================ |
| 432 | // wxCharacterSet and derived classes |
| 433 | // ============================================================================ |
| 434 | |
| 435 | // ---------------------------------------------------------------------------- |
| 436 | // wxCharacterSet is the ABC for the classes below |
| 437 | // ---------------------------------------------------------------------------- |
| 438 | |
| 439 | class wxCharacterSet |
| 440 | { |
| 441 | public: |
| 442 | wxCharacterSet(const wxChar*name) : cname(name) {} |
| 443 | virtual ~wxCharacterSet() {} |
| 444 | virtual size_t MB2WC(wchar_t *buf, const char *psz, size_t n) = 0; |
| 445 | virtual size_t WC2MB(char *buf, const wchar_t *psz, size_t n) = 0; |
| 446 | virtual bool usable() const = 0; |
| 447 | public: |
| 448 | const wxChar*cname; |
| 449 | }; |
| 450 | |
| 451 | // ---------------------------------------------------------------------------- |
| 452 | // ID_CharSet: implementation of wxCharacterSet using an existing wxMBConv |
| 453 | // ---------------------------------------------------------------------------- |
| 454 | |
| 455 | class ID_CharSet : public wxCharacterSet |
| 456 | { |
| 457 | public: |
| 458 | ID_CharSet(const wxChar *name, wxMBConv *cnv) |
| 459 | : wxCharacterSet(name), work(cnv) {} |
| 460 | |
| 461 | size_t MB2WC(wchar_t *buf, const char *psz, size_t n) |
| 462 | { return work ? work->MB2WC(buf,psz,n) : (size_t)-1; } |
| 463 | |
| 464 | size_t WC2MB(char *buf, const wchar_t *psz, size_t n) |
| 465 | { return work ? work->WC2MB(buf,psz,n) : (size_t)-1; } |
| 466 | |
| 467 | bool usable() const |
| 468 | { return work!=NULL; } |
| 469 | public: |
| 470 | wxMBConv*work; |
| 471 | }; |
| 472 | |
| 473 | |
| 474 | // ============================================================================ |
| 475 | // The classes doing conversion using the iconv_xxx() functions |
| 476 | // ============================================================================ |
| 477 | |
| 478 | #ifdef HAVE_ICONV |
| 479 | |
| 480 | // VS: glibc 2.1.3 is broken in that iconv() conversion to/from UCS4 fails with E2BIG |
| 481 | // if output buffer is _exactly_ as big as needed. Such case is (unless there's |
| 482 | // yet another bug in glibc) the only case when iconv() returns with (size_t)-1 |
| 483 | // (which means error) and says there are 0 bytes left in the input buffer -- |
| 484 | // when _real_ error occurs, bytes-left-in-input buffer is non-zero. Hence, |
| 485 | // this alternative test for iconv() failure. |
| 486 | // [This bug does not appear in glibc 2.2.] |
| 487 | #if defined(__GLIBC__) && __GLIBC__ == 2 && __GLIBC_MINOR__ <= 1 |
| 488 | #define ICONV_FAILED(cres, bufLeft) ((cres == (size_t)-1) && \ |
| 489 | (errno != E2BIG || bufLeft != 0)) |
| 490 | #else |
| 491 | #define ICONV_FAILED(cres, bufLeft) (cres == (size_t)-1) |
| 492 | #endif |
| 493 | |
| 494 | #define ICONV_CHAR_CAST(x) ((ICONV_CONST char **)(x)) |
| 495 | |
| 496 | // ---------------------------------------------------------------------------- |
| 497 | // IC_CharSet: encapsulates an iconv character set |
| 498 | // ---------------------------------------------------------------------------- |
| 499 | |
| 500 | class IC_CharSet : public wxCharacterSet |
| 501 | { |
| 502 | public: |
| 503 | IC_CharSet(const wxChar *name); |
| 504 | virtual ~IC_CharSet(); |
| 505 | |
| 506 | virtual size_t MB2WC(wchar_t *buf, const char *psz, size_t n); |
| 507 | virtual size_t WC2MB(char *buf, const wchar_t *psz, size_t n); |
| 508 | |
| 509 | bool usable() const |
| 510 | { return (m2w != (iconv_t)-1) && (w2m != (iconv_t)-1); } |
| 511 | |
| 512 | protected: |
| 513 | // the iconv handlers used to translate from multibyte to wide char and in |
| 514 | // the other direction |
| 515 | iconv_t m2w, |
| 516 | w2m; |
| 517 | |
| 518 | private: |
| 519 | // the name (for iconv_open()) of a wide char charset - if none is |
| 520 | // available on this machine, it will remain NULL |
| 521 | static const char *ms_wcCharsetName; |
| 522 | |
| 523 | // true if the wide char encoding we use (i.e. ms_wcCharsetName) has |
| 524 | // different endian-ness than the native one |
| 525 | static bool ms_wcNeedsSwap; |
| 526 | }; |
| 527 | |
| 528 | const char *IC_CharSet::ms_wcCharsetName = NULL; |
| 529 | bool IC_CharSet::ms_wcNeedsSwap = FALSE; |
| 530 | |
| 531 | IC_CharSet::IC_CharSet(const wxChar *name) |
| 532 | : wxCharacterSet(name) |
| 533 | { |
| 534 | // check for charset that represents wchar_t: |
| 535 | if (ms_wcCharsetName == NULL) |
| 536 | { |
| 537 | ms_wcNeedsSwap = FALSE; |
| 538 | |
| 539 | // try charset with explicit bytesex info (e.g. "UCS-4LE"): |
| 540 | ms_wcCharsetName = WC_NAME_BEST; |
| 541 | m2w = iconv_open(ms_wcCharsetName, wxConvLibc.cWX2MB(name)); |
| 542 | |
| 543 | if (m2w == (iconv_t)-1) |
| 544 | { |
| 545 | // try charset w/o bytesex info (e.g. "UCS4") |
| 546 | // and check for bytesex ourselves: |
| 547 | ms_wcCharsetName = WC_NAME; |
| 548 | m2w = iconv_open(ms_wcCharsetName, wxConvLibc.cWX2MB(name)); |
| 549 | |
| 550 | // last bet, try if it knows WCHAR_T pseudo-charset |
| 551 | if (m2w == (iconv_t)-1) |
| 552 | { |
| 553 | ms_wcCharsetName = "WCHAR_T"; |
| 554 | m2w = iconv_open(ms_wcCharsetName, wxConvLibc.cWX2MB(name)); |
| 555 | } |
| 556 | |
| 557 | if (m2w != (iconv_t)-1) |
| 558 | { |
| 559 | char buf[2], *bufPtr; |
| 560 | wchar_t wbuf[2], *wbufPtr; |
| 561 | size_t insz, outsz; |
| 562 | size_t res; |
| 563 | |
| 564 | buf[0] = 'A'; |
| 565 | buf[1] = 0; |
| 566 | wbuf[0] = 0; |
| 567 | insz = 2; |
| 568 | outsz = SIZEOF_WCHAR_T * 2; |
| 569 | wbufPtr = wbuf; |
| 570 | bufPtr = buf; |
| 571 | |
| 572 | res = iconv(m2w, ICONV_CHAR_CAST(&bufPtr), &insz, |
| 573 | (char**)&wbufPtr, &outsz); |
| 574 | |
| 575 | if (ICONV_FAILED(res, insz)) |
| 576 | { |
| 577 | ms_wcCharsetName = NULL; |
| 578 | wxLogLastError(wxT("iconv")); |
| 579 | wxLogError(_("Convertion to charset '%s' doesn't work."), name); |
| 580 | } |
| 581 | else |
| 582 | { |
| 583 | ms_wcNeedsSwap = wbuf[0] != (wchar_t)buf[0]; |
| 584 | } |
| 585 | } |
| 586 | else |
| 587 | { |
| 588 | ms_wcCharsetName = NULL; |
| 589 | |
| 590 | // VS: we must not output an error here, since wxWindows will safely |
| 591 | // fall back to using wxEncodingConverter. |
| 592 | wxLogTrace(wxT("strconv"), wxT("Impossible to convert to/from charset '%s' with iconv, falling back to wxEncodingConverter."), name); |
| 593 | //wxLogError( |
| 594 | } |
| 595 | } |
| 596 | wxLogTrace(wxT("strconv"), wxT("wchar_t charset is '%s', needs swap: %i"), ms_wcCharsetName, ms_wcNeedsSwap); |
| 597 | } |
| 598 | else // we already have ms_wcCharsetName |
| 599 | { |
| 600 | m2w = iconv_open(ms_wcCharsetName, wxConvLibc.cWX2MB(name)); |
| 601 | } |
| 602 | |
| 603 | // NB: don't ever pass NULL to iconv_open(), it may crash! |
| 604 | if ( ms_wcCharsetName ) |
| 605 | { |
| 606 | w2m = iconv_open(wxConvLibc.cWX2MB(name), ms_wcCharsetName); |
| 607 | } |
| 608 | else |
| 609 | { |
| 610 | w2m = (iconv_t)-1; |
| 611 | } |
| 612 | } |
| 613 | |
| 614 | IC_CharSet::~IC_CharSet() |
| 615 | { |
| 616 | if ( m2w != (iconv_t)-1 ) |
| 617 | iconv_close(m2w); |
| 618 | if ( w2m != (iconv_t)-1 ) |
| 619 | iconv_close(w2m); |
| 620 | } |
| 621 | |
| 622 | size_t IC_CharSet::MB2WC(wchar_t *buf, const char *psz, size_t n) |
| 623 | { |
| 624 | size_t inbuf = strlen(psz); |
| 625 | size_t outbuf = n * SIZEOF_WCHAR_T; |
| 626 | size_t res, cres; |
| 627 | // VS: Use these instead of psz, buf because iconv() modifies its arguments: |
| 628 | wchar_t *bufPtr = buf; |
| 629 | const char *pszPtr = psz; |
| 630 | |
| 631 | if (buf) |
| 632 | { |
| 633 | // have destination buffer, convert there |
| 634 | cres = iconv(m2w, |
| 635 | ICONV_CHAR_CAST(&pszPtr), &inbuf, |
| 636 | (char**)&bufPtr, &outbuf); |
| 637 | res = n - (outbuf / SIZEOF_WCHAR_T); |
| 638 | |
| 639 | if (ms_wcNeedsSwap) |
| 640 | { |
| 641 | // convert to native endianness |
| 642 | WC_BSWAP(buf /* _not_ bufPtr */, res) |
| 643 | } |
| 644 | } |
| 645 | else |
| 646 | { |
| 647 | // no destination buffer... convert using temp buffer |
| 648 | // to calculate destination buffer requirement |
| 649 | wchar_t tbuf[8]; |
| 650 | res = 0; |
| 651 | do { |
| 652 | bufPtr = tbuf; |
| 653 | outbuf = 8*SIZEOF_WCHAR_T; |
| 654 | |
| 655 | cres = iconv(m2w, |
| 656 | ICONV_CHAR_CAST(&pszPtr), &inbuf, |
| 657 | (char**)&bufPtr, &outbuf ); |
| 658 | |
| 659 | res += 8-(outbuf/SIZEOF_WCHAR_T); |
| 660 | } while ((cres==(size_t)-1) && (errno==E2BIG)); |
| 661 | } |
| 662 | |
| 663 | if (ICONV_FAILED(cres, inbuf)) |
| 664 | { |
| 665 | //VS: it is ok if iconv fails, hence trace only |
| 666 | wxLogTrace(wxT("strconv"), wxT("iconv failed: %s"), wxSysErrorMsg(wxSysErrorCode())); |
| 667 | return (size_t)-1; |
| 668 | } |
| 669 | |
| 670 | return res; |
| 671 | } |
| 672 | |
| 673 | size_t IC_CharSet::WC2MB(char *buf, const wchar_t *psz, size_t n) |
| 674 | { |
| 675 | #if defined(__BORLANDC__) && (__BORLANDC__ > 0x530) |
| 676 | size_t inbuf = std::wcslen(psz) * SIZEOF_WCHAR_T; |
| 677 | #else |
| 678 | size_t inbuf = ::wcslen(psz) * SIZEOF_WCHAR_T; |
| 679 | #endif |
| 680 | size_t outbuf = n; |
| 681 | size_t res, cres; |
| 682 | |
| 683 | wchar_t *tmpbuf = 0; |
| 684 | |
| 685 | if (ms_wcNeedsSwap) |
| 686 | { |
| 687 | // need to copy to temp buffer to switch endianness |
| 688 | // this absolutely doesn't rock! |
| 689 | // (no, doing WC_BSWAP twice on the original buffer won't help, as it |
| 690 | // could be in read-only memory, or be accessed in some other thread) |
| 691 | tmpbuf=(wchar_t*)malloc((inbuf+1)*SIZEOF_WCHAR_T); |
| 692 | memcpy(tmpbuf,psz,(inbuf+1)*SIZEOF_WCHAR_T); |
| 693 | WC_BSWAP(tmpbuf, inbuf) |
| 694 | psz=tmpbuf; |
| 695 | } |
| 696 | |
| 697 | if (buf) |
| 698 | { |
| 699 | // have destination buffer, convert there |
| 700 | cres = iconv( w2m, ICONV_CHAR_CAST(&psz), &inbuf, &buf, &outbuf ); |
| 701 | |
| 702 | res = n-outbuf; |
| 703 | } |
| 704 | else |
| 705 | { |
| 706 | // no destination buffer... convert using temp buffer |
| 707 | // to calculate destination buffer requirement |
| 708 | char tbuf[16]; |
| 709 | res = 0; |
| 710 | do { |
| 711 | buf = tbuf; outbuf = 16; |
| 712 | |
| 713 | cres = iconv( w2m, ICONV_CHAR_CAST(&psz), &inbuf, &buf, &outbuf ); |
| 714 | |
| 715 | res += 16 - outbuf; |
| 716 | } while ((cres==(size_t)-1) && (errno==E2BIG)); |
| 717 | } |
| 718 | |
| 719 | if (ms_wcNeedsSwap) |
| 720 | { |
| 721 | free(tmpbuf); |
| 722 | } |
| 723 | |
| 724 | if (ICONV_FAILED(cres, inbuf)) |
| 725 | { |
| 726 | //VS: it is ok if iconv fails, hence trace only |
| 727 | wxLogTrace(wxT("strconv"), wxT("iconv failed: %s"), wxSysErrorMsg(wxSysErrorCode())); |
| 728 | return (size_t)-1; |
| 729 | } |
| 730 | |
| 731 | return res; |
| 732 | } |
| 733 | |
| 734 | #endif // HAVE_ICONV |
| 735 | |
| 736 | // ============================================================================ |
| 737 | // Win32 conversion classes |
| 738 | // ============================================================================ |
| 739 | |
| 740 | #if defined(__WIN32__) && !defined(__WXMICROWIN__) |
| 741 | |
| 742 | extern long wxCharsetToCodepage(const wxChar *charset); // from utils.cpp |
| 743 | |
| 744 | class CP_CharSet : public wxCharacterSet |
| 745 | { |
| 746 | public: |
| 747 | CP_CharSet(const wxChar* name) |
| 748 | : wxCharacterSet(name) |
| 749 | { |
| 750 | m_CodePage = wxCharsetToCodepage(name); |
| 751 | } |
| 752 | |
| 753 | size_t MB2WC(wchar_t *buf, const char *psz, size_t n) |
| 754 | { |
| 755 | size_t len = |
| 756 | MultiByteToWideChar(m_CodePage, 0, psz, -1, buf, buf ? n : 0); |
| 757 | //VS: returns # of written chars for buf!=NULL and *size* |
| 758 | // needed buffer for buf==NULL |
| 759 | return len ? (buf ? len : len-1) : (size_t)-1; |
| 760 | } |
| 761 | |
| 762 | size_t WC2MB(char *buf, const wchar_t *psz, size_t n) |
| 763 | { |
| 764 | size_t len = WideCharToMultiByte(m_CodePage, 0, psz, -1, buf, |
| 765 | buf ? n : 0, NULL, NULL); |
| 766 | //VS: returns # of written chars for buf!=NULL and *size* |
| 767 | // needed buffer for buf==NULL |
| 768 | return len ? (buf ? len : len-1) : (size_t)-1; |
| 769 | } |
| 770 | |
| 771 | bool usable() const |
| 772 | { return m_CodePage != -1; } |
| 773 | |
| 774 | public: |
| 775 | long m_CodePage; |
| 776 | }; |
| 777 | #endif // __WIN32__ |
| 778 | |
| 779 | // ============================================================================ |
| 780 | // wxEncodingConverter based conversion classes |
| 781 | // ============================================================================ |
| 782 | |
| 783 | #if wxUSE_FONTMAP |
| 784 | |
| 785 | class EC_CharSet : public wxCharacterSet |
| 786 | { |
| 787 | public: |
| 788 | // temporarily just use wxEncodingConverter stuff, |
| 789 | // so that it works while a better implementation is built |
| 790 | EC_CharSet(const wxChar* name) : wxCharacterSet(name), |
| 791 | enc(wxFONTENCODING_SYSTEM) |
| 792 | { |
| 793 | if (name) |
| 794 | enc = wxFontMapper::Get()->CharsetToEncoding(name, FALSE); |
| 795 | |
| 796 | m_ok = m2w.Init(enc, wxFONTENCODING_UNICODE) && |
| 797 | w2m.Init(wxFONTENCODING_UNICODE, enc); |
| 798 | } |
| 799 | |
| 800 | size_t MB2WC(wchar_t *buf, const char *psz, size_t WXUNUSED(n)) |
| 801 | { |
| 802 | size_t inbuf = strlen(psz); |
| 803 | if (buf) |
| 804 | m2w.Convert(psz,buf); |
| 805 | return inbuf; |
| 806 | } |
| 807 | |
| 808 | size_t WC2MB(char *buf, const wchar_t *psz, size_t WXUNUSED(n)) |
| 809 | { |
| 810 | #if ( defined(__BORLANDC__) && (__BORLANDC__ > 0x530) ) \ |
| 811 | || ( defined(__MWERKS__) && defined(__WXMSW__) ) |
| 812 | size_t inbuf = std::wcslen(psz); |
| 813 | #else |
| 814 | size_t inbuf = ::wcslen(psz); |
| 815 | #endif |
| 816 | if (buf) |
| 817 | w2m.Convert(psz,buf); |
| 818 | |
| 819 | return inbuf; |
| 820 | } |
| 821 | |
| 822 | bool usable() const { return m_ok; } |
| 823 | |
| 824 | public: |
| 825 | wxFontEncoding enc; |
| 826 | wxEncodingConverter m2w, w2m; |
| 827 | |
| 828 | // were we initialized successfully? |
| 829 | bool m_ok; |
| 830 | }; |
| 831 | |
| 832 | #endif // wxUSE_FONTMAP |
| 833 | |
| 834 | // ---------------------------------------------------------------------------- |
| 835 | // the function creating the wxCharacterSet for the specified charset on the |
| 836 | // current system, trying all possibilities |
| 837 | // ---------------------------------------------------------------------------- |
| 838 | |
| 839 | static wxCharacterSet *wxGetCharacterSet(const wxChar *name) |
| 840 | { |
| 841 | // check for the special case of ASCII charset |
| 842 | #if wxUSE_FONTMAP |
| 843 | if ( wxFontMapper::Get()->CharsetToEncoding(name) == wxFONTENCODING_DEFAULT ) |
| 844 | #else // wxUSE_FONTMAP |
| 845 | if ( !name ) |
| 846 | #endif // wxUSE_FONTMAP/!wxUSE_FONTMAP |
| 847 | { |
| 848 | // don't convert at all |
| 849 | return NULL; |
| 850 | } |
| 851 | |
| 852 | // the test above must have taken care of this case |
| 853 | wxCHECK_MSG( name, NULL, _T("NULL name must be wxFONTENCODING_DEFAULT") ); |
| 854 | |
| 855 | wxCharacterSet *cset; |
| 856 | |
| 857 | if ( wxStricmp(name, wxT("UTF8")) == 0 || wxStricmp(name, wxT("UTF-8")) == 0) |
| 858 | { |
| 859 | cset = new ID_CharSet(name, &wxConvUTF8); |
| 860 | } |
| 861 | else |
| 862 | { |
| 863 | #ifdef HAVE_ICONV |
| 864 | cset = new IC_CharSet(name); |
| 865 | #else // !HAVE_ICONV |
| 866 | cset = NULL; |
| 867 | #endif // HAVE_ICONV/!HAVE_ICONV |
| 868 | } |
| 869 | |
| 870 | // it can only be NULL in this case |
| 871 | #ifndef HAVE_ICONV |
| 872 | if ( cset ) |
| 873 | #endif // !HAVE_ICONV |
| 874 | { |
| 875 | if ( cset->usable() ) |
| 876 | return cset; |
| 877 | |
| 878 | delete cset; |
| 879 | cset = NULL; |
| 880 | } |
| 881 | |
| 882 | #if defined(__WIN32__) && !defined(__WXMICROWIN__) |
| 883 | cset = new CP_CharSet(name); |
| 884 | if ( cset->usable() ) |
| 885 | return cset; |
| 886 | |
| 887 | delete cset; |
| 888 | cset = NULL; |
| 889 | #endif // __WIN32__ |
| 890 | |
| 891 | #if wxUSE_FONTMAP |
| 892 | cset = new EC_CharSet(name); |
| 893 | if ( cset->usable() ) |
| 894 | return cset; |
| 895 | |
| 896 | delete cset; |
| 897 | cset = NULL; |
| 898 | #endif // wxUSE_FONTMAP |
| 899 | |
| 900 | wxLogError(_("Cannot convert from encoding '%s'!"), name); |
| 901 | |
| 902 | return NULL; |
| 903 | } |
| 904 | |
| 905 | // ============================================================================ |
| 906 | // wxCSConv implementation |
| 907 | // ============================================================================ |
| 908 | |
| 909 | wxCSConv::wxCSConv(const wxChar *charset) |
| 910 | { |
| 911 | m_name = (wxChar *)NULL; |
| 912 | m_cset = (wxCharacterSet *) NULL; |
| 913 | m_deferred = TRUE; |
| 914 | |
| 915 | SetName(charset); |
| 916 | } |
| 917 | |
| 918 | wxCSConv::~wxCSConv() |
| 919 | { |
| 920 | Clear(); |
| 921 | } |
| 922 | |
| 923 | void wxCSConv::Clear() |
| 924 | { |
| 925 | if (m_name) |
| 926 | free(m_name); |
| 927 | if (m_cset) |
| 928 | delete m_cset; |
| 929 | m_name = NULL; |
| 930 | m_cset = NULL; |
| 931 | } |
| 932 | |
| 933 | void wxCSConv::SetName(const wxChar *charset) |
| 934 | { |
| 935 | if (charset) |
| 936 | { |
| 937 | m_name = wxStrdup(charset); |
| 938 | m_deferred = TRUE; |
| 939 | } |
| 940 | } |
| 941 | |
| 942 | void wxCSConv::LoadNow() |
| 943 | { |
| 944 | if (m_deferred) |
| 945 | { |
| 946 | if ( !m_name ) |
| 947 | { |
| 948 | wxString name = wxLocale::GetSystemEncodingName(); |
| 949 | if ( !name.empty() ) |
| 950 | SetName(name); |
| 951 | } |
| 952 | |
| 953 | // wxGetCharacterSet() complains about NULL name |
| 954 | m_cset = m_name ? wxGetCharacterSet(m_name) : NULL; |
| 955 | m_deferred = FALSE; |
| 956 | } |
| 957 | } |
| 958 | |
| 959 | size_t wxCSConv::MB2WC(wchar_t *buf, const char *psz, size_t n) const |
| 960 | { |
| 961 | ((wxCSConv *)this)->LoadNow(); // discard constness |
| 962 | |
| 963 | if (m_cset) |
| 964 | return m_cset->MB2WC(buf, psz, n); |
| 965 | |
| 966 | // latin-1 (direct) |
| 967 | size_t len = strlen(psz); |
| 968 | |
| 969 | if (buf) |
| 970 | { |
| 971 | for (size_t c = 0; c <= len; c++) |
| 972 | buf[c] = (unsigned char)(psz[c]); |
| 973 | } |
| 974 | |
| 975 | return len; |
| 976 | } |
| 977 | |
| 978 | size_t wxCSConv::WC2MB(char *buf, const wchar_t *psz, size_t n) const |
| 979 | { |
| 980 | ((wxCSConv *)this)->LoadNow(); // discard constness |
| 981 | |
| 982 | if (m_cset) |
| 983 | return m_cset->WC2MB(buf, psz, n); |
| 984 | |
| 985 | // latin-1 (direct) |
| 986 | #if ( defined(__BORLANDC__) && (__BORLANDC__ > 0x530) ) \ |
| 987 | || ( defined(__MWERKS__) && defined(__WXMSW__) ) |
| 988 | size_t len=std::wcslen(psz); |
| 989 | #else |
| 990 | size_t len=::wcslen(psz); |
| 991 | #endif |
| 992 | if (buf) |
| 993 | { |
| 994 | for (size_t c = 0; c <= len; c++) |
| 995 | buf[c] = (psz[c] > 0xff) ? '?' : psz[c]; |
| 996 | } |
| 997 | |
| 998 | return len; |
| 999 | } |
| 1000 | |
| 1001 | #endif // wxUSE_WCHAR_T |
| 1002 | |
| 1003 | |