| 1 | ///////////////////////////////////////////////////////////////////////////// |
| 2 | // Name: src/common/strconv.cpp |
| 3 | // Purpose: Unicode conversion classes |
| 4 | // Author: Ove Kaaven, Robert Roebling, Vadim Zeitlin, Vaclav Slavik, |
| 5 | // Ryan Norton, Fredrik Roubert (UTF7) |
| 6 | // Modified by: |
| 7 | // Created: 29/01/98 |
| 8 | // RCS-ID: $Id$ |
| 9 | // Copyright: (c) 1999 Ove Kaaven, Robert Roebling, Vaclav Slavik |
| 10 | // (c) 2000-2003 Vadim Zeitlin |
| 11 | // (c) 2004 Ryan Norton, Fredrik Roubert |
| 12 | // Licence: wxWindows licence |
| 13 | ///////////////////////////////////////////////////////////////////////////// |
| 14 | |
| 15 | // For compilers that support precompilation, includes "wx.h". |
| 16 | #include "wx/wxprec.h" |
| 17 | |
| 18 | #ifdef __BORLANDC__ |
| 19 | #pragma hdrstop |
| 20 | #endif //__BORLANDC__ |
| 21 | |
| 22 | #ifndef WX_PRECOMP |
| 23 | #include "wx/intl.h" |
| 24 | #include "wx/log.h" |
| 25 | #include "wx/utils.h" |
| 26 | #include "wx/hashmap.h" |
| 27 | #endif |
| 28 | |
| 29 | #include "wx/strconv.h" |
| 30 | |
| 31 | #ifndef __WXWINCE__ |
| 32 | #include <errno.h> |
| 33 | #endif |
| 34 | |
| 35 | #include <ctype.h> |
| 36 | #include <string.h> |
| 37 | #include <stdlib.h> |
| 38 | |
| 39 | #if defined(__WIN32__) && !defined(__WXMICROWIN__) |
| 40 | #include "wx/msw/private.h" |
| 41 | #include "wx/msw/missing.h" |
| 42 | #define wxHAVE_WIN32_MB2WC |
| 43 | #endif |
| 44 | |
| 45 | #ifdef HAVE_ICONV |
| 46 | #include <iconv.h> |
| 47 | #include "wx/thread.h" |
| 48 | #endif |
| 49 | |
| 50 | #include "wx/encconv.h" |
| 51 | #include "wx/fontmap.h" |
| 52 | |
| 53 | #ifdef __DARWIN__ |
| 54 | #include "wx/osx/core/private/strconv_cf.h" |
| 55 | #endif //def __DARWIN__ |
| 56 | |
| 57 | |
| 58 | #define TRACE_STRCONV wxT("strconv") |
| 59 | |
| 60 | // WC_UTF16 is defined only if sizeof(wchar_t) == 2, otherwise it's supposed to |
| 61 | // be 4 bytes |
| 62 | #if SIZEOF_WCHAR_T == 2 |
| 63 | #define WC_UTF16 |
| 64 | #endif |
| 65 | |
| 66 | |
| 67 | // ============================================================================ |
| 68 | // implementation |
| 69 | // ============================================================================ |
| 70 | |
| 71 | // helper function of cMB2WC(): check if n bytes at this location are all NUL |
| 72 | static bool NotAllNULs(const char *p, size_t n) |
| 73 | { |
| 74 | while ( n && *p++ == '\0' ) |
| 75 | n--; |
| 76 | |
| 77 | return n != 0; |
| 78 | } |
| 79 | |
| 80 | // ---------------------------------------------------------------------------- |
| 81 | // UTF-16 en/decoding to/from UCS-4 with surrogates handling |
| 82 | // ---------------------------------------------------------------------------- |
| 83 | |
| 84 | static size_t encode_utf16(wxUint32 input, wxUint16 *output) |
| 85 | { |
| 86 | if (input <= 0xffff) |
| 87 | { |
| 88 | if (output) |
| 89 | *output = (wxUint16) input; |
| 90 | |
| 91 | return 1; |
| 92 | } |
| 93 | else if (input >= 0x110000) |
| 94 | { |
| 95 | return wxCONV_FAILED; |
| 96 | } |
| 97 | else |
| 98 | { |
| 99 | if (output) |
| 100 | { |
| 101 | *output++ = (wxUint16) ((input >> 10) + 0xd7c0); |
| 102 | *output = (wxUint16) ((input & 0x3ff) + 0xdc00); |
| 103 | } |
| 104 | |
| 105 | return 2; |
| 106 | } |
| 107 | } |
| 108 | |
| 109 | static size_t decode_utf16(const wxUint16* input, wxUint32& output) |
| 110 | { |
| 111 | if ((*input < 0xd800) || (*input > 0xdfff)) |
| 112 | { |
| 113 | output = *input; |
| 114 | return 1; |
| 115 | } |
| 116 | else if ((input[1] < 0xdc00) || (input[1] > 0xdfff)) |
| 117 | { |
| 118 | output = *input; |
| 119 | return wxCONV_FAILED; |
| 120 | } |
| 121 | else |
| 122 | { |
| 123 | output = ((input[0] - 0xd7c0) << 10) + (input[1] - 0xdc00); |
| 124 | return 2; |
| 125 | } |
| 126 | } |
| 127 | |
| 128 | #ifdef WC_UTF16 |
| 129 | typedef wchar_t wxDecodeSurrogate_t; |
| 130 | #else // !WC_UTF16 |
| 131 | typedef wxUint16 wxDecodeSurrogate_t; |
| 132 | #endif // WC_UTF16/!WC_UTF16 |
| 133 | |
| 134 | // returns the next UTF-32 character from the wchar_t buffer and advances the |
| 135 | // pointer to the character after this one |
| 136 | // |
| 137 | // if an invalid character is found, *pSrc is set to NULL, the caller must |
| 138 | // check for this |
| 139 | static wxUint32 wxDecodeSurrogate(const wxDecodeSurrogate_t **pSrc) |
| 140 | { |
| 141 | wxUint32 out; |
| 142 | const size_t |
| 143 | n = decode_utf16(reinterpret_cast<const wxUint16 *>(*pSrc), out); |
| 144 | if ( n == wxCONV_FAILED ) |
| 145 | *pSrc = NULL; |
| 146 | else |
| 147 | *pSrc += n; |
| 148 | |
| 149 | return out; |
| 150 | } |
| 151 | |
| 152 | // ---------------------------------------------------------------------------- |
| 153 | // wxMBConv |
| 154 | // ---------------------------------------------------------------------------- |
| 155 | |
| 156 | size_t |
| 157 | wxMBConv::ToWChar(wchar_t *dst, size_t dstLen, |
| 158 | const char *src, size_t srcLen) const |
| 159 | { |
| 160 | // although new conversion classes are supposed to implement this function |
| 161 | // directly, the existing ones only implement the old MB2WC() and so, to |
| 162 | // avoid to have to rewrite all conversion classes at once, we provide a |
| 163 | // default (but not efficient) implementation of this one in terms of the |
| 164 | // old function by copying the input to ensure that it's NUL-terminated and |
| 165 | // then using MB2WC() to convert it |
| 166 | // |
| 167 | // moreover, some conversion classes simply can't implement ToWChar() |
| 168 | // directly, the primary example is wxConvLibc: mbstowcs() only handles |
| 169 | // NUL-terminated strings |
| 170 | |
| 171 | // the number of chars [which would be] written to dst [if it were not NULL] |
| 172 | size_t dstWritten = 0; |
| 173 | |
| 174 | // the number of NULs terminating this string |
| 175 | size_t nulLen = 0; // not really needed, but just to avoid warnings |
| 176 | |
| 177 | // if we were not given the input size we just have to assume that the |
| 178 | // string is properly terminated as we have no way of knowing how long it |
| 179 | // is anyhow, but if we do have the size check whether there are enough |
| 180 | // NULs at the end |
| 181 | wxCharBuffer bufTmp; |
| 182 | const char *srcEnd; |
| 183 | if ( srcLen != wxNO_LEN ) |
| 184 | { |
| 185 | // we need to know how to find the end of this string |
| 186 | nulLen = GetMBNulLen(); |
| 187 | if ( nulLen == wxCONV_FAILED ) |
| 188 | return wxCONV_FAILED; |
| 189 | |
| 190 | // if there are enough NULs we can avoid the copy |
| 191 | if ( srcLen < nulLen || NotAllNULs(src + srcLen - nulLen, nulLen) ) |
| 192 | { |
| 193 | // make a copy in order to properly NUL-terminate the string |
| 194 | bufTmp = wxCharBuffer(srcLen + nulLen - 1 /* 1 will be added */); |
| 195 | char * const p = bufTmp.data(); |
| 196 | memcpy(p, src, srcLen); |
| 197 | for ( char *s = p + srcLen; s < p + srcLen + nulLen; s++ ) |
| 198 | *s = '\0'; |
| 199 | |
| 200 | src = bufTmp; |
| 201 | } |
| 202 | |
| 203 | srcEnd = src + srcLen; |
| 204 | } |
| 205 | else // quit after the first loop iteration |
| 206 | { |
| 207 | srcEnd = NULL; |
| 208 | } |
| 209 | |
| 210 | // the idea of this code is straightforward: it converts a NUL-terminated |
| 211 | // chunk of the string during each iteration and updates the output buffer |
| 212 | // with the result |
| 213 | // |
| 214 | // all the complication come from the fact that this function, for |
| 215 | // historical reasons, must behave in 2 subtly different ways when it's |
| 216 | // called with a fixed number of characters and when it's called for the |
| 217 | // entire NUL-terminated string: in the former case (srcEnd != NULL) we |
| 218 | // must count all characters we convert, NUL or not; but in the latter we |
| 219 | // do not count the trailing NUL -- but still count all the NULs inside the |
| 220 | // string |
| 221 | // |
| 222 | // so for the (simple) former case we just always count the trailing NUL, |
| 223 | // but for the latter we need to wait until we see if there is going to be |
| 224 | // another loop iteration and only count it then |
| 225 | for ( ;; ) |
| 226 | { |
| 227 | // try to convert the current chunk |
| 228 | size_t lenChunk = MB2WC(NULL, src, 0); |
| 229 | if ( lenChunk == wxCONV_FAILED ) |
| 230 | return wxCONV_FAILED; |
| 231 | |
| 232 | dstWritten += lenChunk; |
| 233 | if ( !srcEnd ) |
| 234 | dstWritten++; |
| 235 | |
| 236 | if ( !lenChunk ) |
| 237 | { |
| 238 | // nothing left in the input string, conversion succeeded |
| 239 | break; |
| 240 | } |
| 241 | |
| 242 | if ( dst ) |
| 243 | { |
| 244 | if ( dstWritten > dstLen ) |
| 245 | return wxCONV_FAILED; |
| 246 | |
| 247 | // +1 is for trailing NUL |
| 248 | if ( MB2WC(dst, src, lenChunk + 1) == wxCONV_FAILED ) |
| 249 | return wxCONV_FAILED; |
| 250 | |
| 251 | dst += lenChunk; |
| 252 | if ( !srcEnd ) |
| 253 | dst++; |
| 254 | } |
| 255 | |
| 256 | if ( !srcEnd ) |
| 257 | { |
| 258 | // we convert just one chunk in this case as this is the entire |
| 259 | // string anyhow (and we don't count the trailing NUL in this case) |
| 260 | break; |
| 261 | } |
| 262 | |
| 263 | // advance the input pointer past the end of this chunk: notice that we |
| 264 | // will always stop before srcEnd because we know that the chunk is |
| 265 | // always properly NUL-terminated |
| 266 | while ( NotAllNULs(src, nulLen) ) |
| 267 | { |
| 268 | // notice that we must skip over multiple bytes here as we suppose |
| 269 | // that if NUL takes 2 or 4 bytes, then all the other characters do |
| 270 | // too and so if advanced by a single byte we might erroneously |
| 271 | // detect sequences of NUL bytes in the middle of the input |
| 272 | src += nulLen; |
| 273 | } |
| 274 | |
| 275 | // if the buffer ends before this NUL, we shouldn't count it in our |
| 276 | // output so skip the code below |
| 277 | if ( src == srcEnd ) |
| 278 | break; |
| 279 | |
| 280 | // do count this terminator as it's inside the buffer we convert |
| 281 | dstWritten++; |
| 282 | if ( dst ) |
| 283 | dst++; |
| 284 | |
| 285 | src += nulLen; // skip the terminator itself |
| 286 | |
| 287 | if ( src >= srcEnd ) |
| 288 | break; |
| 289 | } |
| 290 | |
| 291 | return dstWritten; |
| 292 | } |
| 293 | |
| 294 | size_t |
| 295 | wxMBConv::FromWChar(char *dst, size_t dstLen, |
| 296 | const wchar_t *src, size_t srcLen) const |
| 297 | { |
| 298 | // the number of chars [which would be] written to dst [if it were not NULL] |
| 299 | size_t dstWritten = 0; |
| 300 | |
| 301 | // if we don't know its length we have no choice but to assume that it is |
| 302 | // NUL-terminated (notice that it can still be NUL-terminated even if |
| 303 | // explicit length is given but it doesn't change our return value) |
| 304 | const bool isNulTerminated = srcLen == wxNO_LEN; |
| 305 | |
| 306 | // make a copy of the input string unless it is already properly |
| 307 | // NUL-terminated |
| 308 | wxWCharBuffer bufTmp; |
| 309 | if ( isNulTerminated ) |
| 310 | { |
| 311 | srcLen = wxWcslen(src) + 1; |
| 312 | } |
| 313 | else if ( srcLen != 0 && src[srcLen - 1] != L'\0' ) |
| 314 | { |
| 315 | // make a copy in order to properly NUL-terminate the string |
| 316 | bufTmp = wxWCharBuffer(srcLen); |
| 317 | memcpy(bufTmp.data(), src, srcLen * sizeof(wchar_t)); |
| 318 | src = bufTmp; |
| 319 | } |
| 320 | |
| 321 | const size_t lenNul = GetMBNulLen(); |
| 322 | for ( const wchar_t * const srcEnd = src + srcLen; |
| 323 | src < srcEnd; |
| 324 | src++ /* skip L'\0' too */ ) |
| 325 | { |
| 326 | // try to convert the current chunk |
| 327 | size_t lenChunk = WC2MB(NULL, src, 0); |
| 328 | if ( lenChunk == wxCONV_FAILED ) |
| 329 | return wxCONV_FAILED; |
| 330 | |
| 331 | dstWritten += lenChunk; |
| 332 | |
| 333 | const wchar_t * const |
| 334 | chunkEnd = isNulTerminated ? srcEnd - 1 : src + wxWcslen(src); |
| 335 | |
| 336 | // our return value accounts for the trailing NUL(s), unlike that of |
| 337 | // WC2MB(), however don't do it for the last NUL we artificially added |
| 338 | // ourselves above |
| 339 | if ( chunkEnd < srcEnd ) |
| 340 | dstWritten += lenNul; |
| 341 | |
| 342 | if ( dst ) |
| 343 | { |
| 344 | if ( dstWritten > dstLen ) |
| 345 | return wxCONV_FAILED; |
| 346 | |
| 347 | // if we know that there is enough space in the destination buffer |
| 348 | // (because we accounted for lenNul in dstWritten above), we can |
| 349 | // convert directly in place -- but otherwise we need another |
| 350 | // temporary buffer to ensure that we don't overwrite the output |
| 351 | wxCharBuffer dstBuf; |
| 352 | char *dstTmp; |
| 353 | if ( chunkEnd == srcEnd ) |
| 354 | { |
| 355 | dstBuf = wxCharBuffer(lenChunk + lenNul - 1); |
| 356 | dstTmp = dstBuf.data(); |
| 357 | } |
| 358 | else |
| 359 | { |
| 360 | dstTmp = dst; |
| 361 | } |
| 362 | |
| 363 | if ( WC2MB(dstTmp, src, lenChunk + lenNul) == wxCONV_FAILED ) |
| 364 | return wxCONV_FAILED; |
| 365 | |
| 366 | if ( dstTmp != dst ) |
| 367 | { |
| 368 | // copy everything up to but excluding the terminating NUL(s) |
| 369 | // into the real output buffer |
| 370 | memcpy(dst, dstTmp, lenChunk); |
| 371 | |
| 372 | // micro-optimization: if dstTmp != dst it means that chunkEnd |
| 373 | // == srcEnd and so we're done, no need to update anything below |
| 374 | break; |
| 375 | } |
| 376 | |
| 377 | dst += lenChunk; |
| 378 | if ( chunkEnd < srcEnd ) |
| 379 | dst += lenNul; |
| 380 | } |
| 381 | |
| 382 | src = chunkEnd; |
| 383 | } |
| 384 | |
| 385 | return dstWritten; |
| 386 | } |
| 387 | |
| 388 | size_t wxMBConv::MB2WC(wchar_t *outBuff, const char *inBuff, size_t outLen) const |
| 389 | { |
| 390 | size_t rc = ToWChar(outBuff, outLen, inBuff); |
| 391 | if ( rc != wxCONV_FAILED ) |
| 392 | { |
| 393 | // ToWChar() returns the buffer length, i.e. including the trailing |
| 394 | // NUL, while this method doesn't take it into account |
| 395 | rc--; |
| 396 | } |
| 397 | |
| 398 | return rc; |
| 399 | } |
| 400 | |
| 401 | size_t wxMBConv::WC2MB(char *outBuff, const wchar_t *inBuff, size_t outLen) const |
| 402 | { |
| 403 | size_t rc = FromWChar(outBuff, outLen, inBuff); |
| 404 | if ( rc != wxCONV_FAILED ) |
| 405 | { |
| 406 | rc -= GetMBNulLen(); |
| 407 | } |
| 408 | |
| 409 | return rc; |
| 410 | } |
| 411 | |
| 412 | wxMBConv::~wxMBConv() |
| 413 | { |
| 414 | // nothing to do here (necessary for Darwin linking probably) |
| 415 | } |
| 416 | |
| 417 | const wxWCharBuffer wxMBConv::cMB2WC(const char *psz) const |
| 418 | { |
| 419 | if ( psz ) |
| 420 | { |
| 421 | // calculate the length of the buffer needed first |
| 422 | const size_t nLen = ToWChar(NULL, 0, psz); |
| 423 | if ( nLen != wxCONV_FAILED ) |
| 424 | { |
| 425 | // now do the actual conversion |
| 426 | wxWCharBuffer buf(nLen - 1 /* +1 added implicitly */); |
| 427 | |
| 428 | // +1 for the trailing NULL |
| 429 | if ( ToWChar(buf.data(), nLen, psz) != wxCONV_FAILED ) |
| 430 | return buf; |
| 431 | } |
| 432 | } |
| 433 | |
| 434 | return wxWCharBuffer(); |
| 435 | } |
| 436 | |
| 437 | const wxCharBuffer wxMBConv::cWC2MB(const wchar_t *pwz) const |
| 438 | { |
| 439 | if ( pwz ) |
| 440 | { |
| 441 | const size_t nLen = FromWChar(NULL, 0, pwz); |
| 442 | if ( nLen != wxCONV_FAILED ) |
| 443 | { |
| 444 | wxCharBuffer buf(nLen - 1); |
| 445 | if ( FromWChar(buf.data(), nLen, pwz) != wxCONV_FAILED ) |
| 446 | return buf; |
| 447 | } |
| 448 | } |
| 449 | |
| 450 | return wxCharBuffer(); |
| 451 | } |
| 452 | |
| 453 | const wxWCharBuffer |
| 454 | wxMBConv::cMB2WC(const char *inBuff, size_t inLen, size_t *outLen) const |
| 455 | { |
| 456 | const size_t dstLen = ToWChar(NULL, 0, inBuff, inLen); |
| 457 | if ( dstLen != wxCONV_FAILED ) |
| 458 | { |
| 459 | // notice that we allocate space for dstLen+1 wide characters here |
| 460 | // because we want the buffer to always be NUL-terminated, even if the |
| 461 | // input isn't (as otherwise the caller has no way to know its length) |
| 462 | wxWCharBuffer wbuf(dstLen); |
| 463 | wbuf.data()[dstLen] = L'\0'; |
| 464 | if ( ToWChar(wbuf.data(), dstLen, inBuff, inLen) != wxCONV_FAILED ) |
| 465 | { |
| 466 | if ( outLen ) |
| 467 | { |
| 468 | *outLen = dstLen; |
| 469 | |
| 470 | // we also need to handle NUL-terminated input strings |
| 471 | // specially: for them the output is the length of the string |
| 472 | // excluding the trailing NUL, however if we're asked to |
| 473 | // convert a specific number of characters we return the length |
| 474 | // of the resulting output even if it's NUL-terminated |
| 475 | if ( inLen == wxNO_LEN ) |
| 476 | (*outLen)--; |
| 477 | } |
| 478 | |
| 479 | return wbuf; |
| 480 | } |
| 481 | } |
| 482 | |
| 483 | if ( outLen ) |
| 484 | *outLen = 0; |
| 485 | |
| 486 | return wxWCharBuffer(); |
| 487 | } |
| 488 | |
| 489 | const wxCharBuffer |
| 490 | wxMBConv::cWC2MB(const wchar_t *inBuff, size_t inLen, size_t *outLen) const |
| 491 | { |
| 492 | size_t dstLen = FromWChar(NULL, 0, inBuff, inLen); |
| 493 | if ( dstLen != wxCONV_FAILED ) |
| 494 | { |
| 495 | const size_t nulLen = GetMBNulLen(); |
| 496 | |
| 497 | // as above, ensure that the buffer is always NUL-terminated, even if |
| 498 | // the input is not |
| 499 | wxCharBuffer buf(dstLen + nulLen - 1); |
| 500 | memset(buf.data() + dstLen, 0, nulLen); |
| 501 | if ( FromWChar(buf.data(), dstLen, inBuff, inLen) != wxCONV_FAILED ) |
| 502 | { |
| 503 | if ( outLen ) |
| 504 | { |
| 505 | *outLen = dstLen; |
| 506 | |
| 507 | if ( inLen == wxNO_LEN ) |
| 508 | { |
| 509 | // in this case both input and output are NUL-terminated |
| 510 | // and we're not supposed to count NUL |
| 511 | *outLen -= nulLen; |
| 512 | } |
| 513 | } |
| 514 | |
| 515 | return buf; |
| 516 | } |
| 517 | } |
| 518 | |
| 519 | if ( outLen ) |
| 520 | *outLen = 0; |
| 521 | |
| 522 | return wxCharBuffer(); |
| 523 | } |
| 524 | |
| 525 | const wxWCharBuffer wxMBConv::cMB2WC(const wxScopedCharBuffer& buf) const |
| 526 | { |
| 527 | const size_t srcLen = buf.length(); |
| 528 | if ( srcLen ) |
| 529 | { |
| 530 | const size_t dstLen = ToWChar(NULL, 0, buf, srcLen); |
| 531 | if ( dstLen != wxCONV_FAILED ) |
| 532 | { |
| 533 | wxWCharBuffer wbuf(dstLen); |
| 534 | wbuf.data()[dstLen] = L'\0'; |
| 535 | if ( ToWChar(wbuf.data(), dstLen, buf, srcLen) != wxCONV_FAILED ) |
| 536 | return wbuf; |
| 537 | } |
| 538 | } |
| 539 | |
| 540 | return wxScopedWCharBuffer::CreateNonOwned(L"", 0); |
| 541 | } |
| 542 | |
| 543 | const wxCharBuffer wxMBConv::cWC2MB(const wxScopedWCharBuffer& wbuf) const |
| 544 | { |
| 545 | const size_t srcLen = wbuf.length(); |
| 546 | if ( srcLen ) |
| 547 | { |
| 548 | const size_t dstLen = FromWChar(NULL, 0, wbuf, srcLen); |
| 549 | if ( dstLen != wxCONV_FAILED ) |
| 550 | { |
| 551 | wxCharBuffer buf(dstLen); |
| 552 | buf.data()[dstLen] = '\0'; |
| 553 | if ( FromWChar(buf.data(), dstLen, wbuf, srcLen) != wxCONV_FAILED ) |
| 554 | return buf; |
| 555 | } |
| 556 | } |
| 557 | |
| 558 | return wxScopedCharBuffer::CreateNonOwned("", 0); |
| 559 | } |
| 560 | |
| 561 | // ---------------------------------------------------------------------------- |
| 562 | // wxMBConvLibc |
| 563 | // ---------------------------------------------------------------------------- |
| 564 | |
| 565 | size_t wxMBConvLibc::MB2WC(wchar_t *buf, const char *psz, size_t n) const |
| 566 | { |
| 567 | return wxMB2WC(buf, psz, n); |
| 568 | } |
| 569 | |
| 570 | size_t wxMBConvLibc::WC2MB(char *buf, const wchar_t *psz, size_t n) const |
| 571 | { |
| 572 | return wxWC2MB(buf, psz, n); |
| 573 | } |
| 574 | |
| 575 | // ---------------------------------------------------------------------------- |
| 576 | // wxConvBrokenFileNames |
| 577 | // ---------------------------------------------------------------------------- |
| 578 | |
| 579 | #ifdef __UNIX__ |
| 580 | |
| 581 | wxConvBrokenFileNames::wxConvBrokenFileNames(const wxString& charset) |
| 582 | { |
| 583 | if ( wxStricmp(charset, wxT("UTF-8")) == 0 || |
| 584 | wxStricmp(charset, wxT("UTF8")) == 0 ) |
| 585 | m_conv = new wxMBConvUTF8(wxMBConvUTF8::MAP_INVALID_UTF8_TO_PUA); |
| 586 | else |
| 587 | m_conv = new wxCSConv(charset); |
| 588 | } |
| 589 | |
| 590 | #endif // __UNIX__ |
| 591 | |
| 592 | // ---------------------------------------------------------------------------- |
| 593 | // UTF-7 |
| 594 | // ---------------------------------------------------------------------------- |
| 595 | |
| 596 | // Implementation (C) 2004 Fredrik Roubert |
| 597 | // |
| 598 | // Changes to work in streaming mode (C) 2008 Vadim Zeitlin |
| 599 | |
| 600 | // |
| 601 | // BASE64 decoding table |
| 602 | // |
| 603 | static const unsigned char utf7unb64[] = |
| 604 | { |
| 605 | 0xff, 0xff, 0xff, 0xff, 0xff, 0xff, 0xff, 0xff, |
| 606 | 0xff, 0xff, 0xff, 0xff, 0xff, 0xff, 0xff, 0xff, |
| 607 | 0xff, 0xff, 0xff, 0xff, 0xff, 0xff, 0xff, 0xff, |
| 608 | 0xff, 0xff, 0xff, 0xff, 0xff, 0xff, 0xff, 0xff, |
| 609 | 0xff, 0xff, 0xff, 0xff, 0xff, 0xff, 0xff, 0xff, |
| 610 | 0xff, 0xff, 0xff, 0x3e, 0xff, 0xff, 0xff, 0x3f, |
| 611 | 0x34, 0x35, 0x36, 0x37, 0x38, 0x39, 0x3a, 0x3b, |
| 612 | 0x3c, 0x3d, 0xff, 0xff, 0xff, 0xff, 0xff, 0xff, |
| 613 | 0xff, 0x00, 0x01, 0x02, 0x03, 0x04, 0x05, 0x06, |
| 614 | 0x07, 0x08, 0x09, 0x0a, 0x0b, 0x0c, 0x0d, 0x0e, |
| 615 | 0x0f, 0x10, 0x11, 0x12, 0x13, 0x14, 0x15, 0x16, |
| 616 | 0x17, 0x18, 0x19, 0xff, 0xff, 0xff, 0xff, 0xff, |
| 617 | 0xff, 0x1a, 0x1b, 0x1c, 0x1d, 0x1e, 0x1f, 0x20, |
| 618 | 0x21, 0x22, 0x23, 0x24, 0x25, 0x26, 0x27, 0x28, |
| 619 | 0x29, 0x2a, 0x2b, 0x2c, 0x2d, 0x2e, 0x2f, 0x30, |
| 620 | 0x31, 0x32, 0x33, 0xff, 0xff, 0xff, 0xff, 0xff, |
| 621 | 0xff, 0xff, 0xff, 0xff, 0xff, 0xff, 0xff, 0xff, |
| 622 | 0xff, 0xff, 0xff, 0xff, 0xff, 0xff, 0xff, 0xff, |
| 623 | 0xff, 0xff, 0xff, 0xff, 0xff, 0xff, 0xff, 0xff, |
| 624 | 0xff, 0xff, 0xff, 0xff, 0xff, 0xff, 0xff, 0xff, |
| 625 | 0xff, 0xff, 0xff, 0xff, 0xff, 0xff, 0xff, 0xff, |
| 626 | 0xff, 0xff, 0xff, 0xff, 0xff, 0xff, 0xff, 0xff, |
| 627 | 0xff, 0xff, 0xff, 0xff, 0xff, 0xff, 0xff, 0xff, |
| 628 | 0xff, 0xff, 0xff, 0xff, 0xff, 0xff, 0xff, 0xff, |
| 629 | 0xff, 0xff, 0xff, 0xff, 0xff, 0xff, 0xff, 0xff, |
| 630 | 0xff, 0xff, 0xff, 0xff, 0xff, 0xff, 0xff, 0xff, |
| 631 | 0xff, 0xff, 0xff, 0xff, 0xff, 0xff, 0xff, 0xff, |
| 632 | 0xff, 0xff, 0xff, 0xff, 0xff, 0xff, 0xff, 0xff, |
| 633 | 0xff, 0xff, 0xff, 0xff, 0xff, 0xff, 0xff, 0xff, |
| 634 | 0xff, 0xff, 0xff, 0xff, 0xff, 0xff, 0xff, 0xff, |
| 635 | 0xff, 0xff, 0xff, 0xff, 0xff, 0xff, 0xff, 0xff, |
| 636 | 0xff, 0xff, 0xff, 0xff, 0xff, 0xff, 0xff, 0xff |
| 637 | }; |
| 638 | |
| 639 | size_t wxMBConvUTF7::ToWChar(wchar_t *dst, size_t dstLen, |
| 640 | const char *src, size_t srcLen) const |
| 641 | { |
| 642 | DecoderState stateOrig, |
| 643 | *statePtr; |
| 644 | if ( srcLen == wxNO_LEN ) |
| 645 | { |
| 646 | // convert the entire string, up to and including the trailing NUL |
| 647 | srcLen = strlen(src) + 1; |
| 648 | |
| 649 | // when working on the entire strings we don't update nor use the shift |
| 650 | // state from the previous call |
| 651 | statePtr = &stateOrig; |
| 652 | } |
| 653 | else // when working with partial strings we do use the shift state |
| 654 | { |
| 655 | statePtr = const_cast<DecoderState *>(&m_stateDecoder); |
| 656 | |
| 657 | // also save the old state to be able to rollback to it on error |
| 658 | stateOrig = m_stateDecoder; |
| 659 | } |
| 660 | |
| 661 | // but to simplify the code below we use this variable in both cases |
| 662 | DecoderState& state = *statePtr; |
| 663 | |
| 664 | |
| 665 | // number of characters [which would have been] written to dst [if it were |
| 666 | // not NULL] |
| 667 | size_t len = 0; |
| 668 | |
| 669 | const char * const srcEnd = src + srcLen; |
| 670 | |
| 671 | while ( (src < srcEnd) && (!dst || (len < dstLen)) ) |
| 672 | { |
| 673 | const unsigned char cc = *src++; |
| 674 | |
| 675 | if ( state.IsShifted() ) |
| 676 | { |
| 677 | const unsigned char dc = utf7unb64[cc]; |
| 678 | if ( dc == 0xff ) |
| 679 | { |
| 680 | // end of encoded part, check that nothing was left: there can |
| 681 | // be up to 4 bits of 0 padding but nothing else (we also need |
| 682 | // to check isLSB as we count bits modulo 8 while a valid UTF-7 |
| 683 | // encoded sequence must contain an integral number of UTF-16 |
| 684 | // characters) |
| 685 | if ( state.isLSB || state.bit > 4 || |
| 686 | (state.accum & ((1 << state.bit) - 1)) ) |
| 687 | { |
| 688 | if ( !len ) |
| 689 | state = stateOrig; |
| 690 | |
| 691 | return wxCONV_FAILED; |
| 692 | } |
| 693 | |
| 694 | state.ToDirect(); |
| 695 | |
| 696 | // re-parse this character normally below unless it's '-' which |
| 697 | // is consumed by the decoder |
| 698 | if ( cc == '-' ) |
| 699 | continue; |
| 700 | } |
| 701 | else // valid encoded character |
| 702 | { |
| 703 | // mini base64 decoder: each character is 6 bits |
| 704 | state.bit += 6; |
| 705 | state.accum <<= 6; |
| 706 | state.accum += dc; |
| 707 | |
| 708 | if ( state.bit >= 8 ) |
| 709 | { |
| 710 | // got the full byte, consume it |
| 711 | state.bit -= 8; |
| 712 | unsigned char b = (state.accum >> state.bit) & 0x00ff; |
| 713 | |
| 714 | if ( state.isLSB ) |
| 715 | { |
| 716 | // we've got the full word, output it |
| 717 | if ( dst ) |
| 718 | *dst++ = (state.msb << 8) | b; |
| 719 | len++; |
| 720 | state.isLSB = false; |
| 721 | } |
| 722 | else // MSB |
| 723 | { |
| 724 | // just store it while we wait for LSB |
| 725 | state.msb = b; |
| 726 | state.isLSB = true; |
| 727 | } |
| 728 | } |
| 729 | } |
| 730 | } |
| 731 | |
| 732 | if ( state.IsDirect() ) |
| 733 | { |
| 734 | // start of an encoded segment? |
| 735 | if ( cc == '+' ) |
| 736 | { |
| 737 | if ( *src == '-' ) |
| 738 | { |
| 739 | // just the encoded plus sign, don't switch to shifted mode |
| 740 | if ( dst ) |
| 741 | *dst++ = '+'; |
| 742 | len++; |
| 743 | src++; |
| 744 | } |
| 745 | else if ( utf7unb64[(unsigned)*src] == 0xff ) |
| 746 | { |
| 747 | // empty encoded chunks are not allowed |
| 748 | if ( !len ) |
| 749 | state = stateOrig; |
| 750 | |
| 751 | return wxCONV_FAILED; |
| 752 | } |
| 753 | else // base-64 encoded chunk follows |
| 754 | { |
| 755 | state.ToShifted(); |
| 756 | } |
| 757 | } |
| 758 | else // not '+' |
| 759 | { |
| 760 | // only printable 7 bit ASCII characters (with the exception of |
| 761 | // NUL, TAB, CR and LF) can be used directly |
| 762 | if ( cc >= 0x7f || (cc < ' ' && |
| 763 | !(cc == '\0' || cc == '\t' || cc == '\r' || cc == '\n')) ) |
| 764 | return wxCONV_FAILED; |
| 765 | |
| 766 | if ( dst ) |
| 767 | *dst++ = cc; |
| 768 | len++; |
| 769 | } |
| 770 | } |
| 771 | } |
| 772 | |
| 773 | if ( !len ) |
| 774 | { |
| 775 | // as we didn't read any characters we should be called with the same |
| 776 | // data (followed by some more new data) again later so don't save our |
| 777 | // state |
| 778 | state = stateOrig; |
| 779 | |
| 780 | return wxCONV_FAILED; |
| 781 | } |
| 782 | |
| 783 | return len; |
| 784 | } |
| 785 | |
| 786 | // |
| 787 | // BASE64 encoding table |
| 788 | // |
| 789 | static const unsigned char utf7enb64[] = |
| 790 | { |
| 791 | 'A', 'B', 'C', 'D', 'E', 'F', 'G', 'H', |
| 792 | 'I', 'J', 'K', 'L', 'M', 'N', 'O', 'P', |
| 793 | 'Q', 'R', 'S', 'T', 'U', 'V', 'W', 'X', |
| 794 | 'Y', 'Z', 'a', 'b', 'c', 'd', 'e', 'f', |
| 795 | 'g', 'h', 'i', 'j', 'k', 'l', 'm', 'n', |
| 796 | 'o', 'p', 'q', 'r', 's', 't', 'u', 'v', |
| 797 | 'w', 'x', 'y', 'z', '0', '1', '2', '3', |
| 798 | '4', '5', '6', '7', '8', '9', '+', '/' |
| 799 | }; |
| 800 | |
| 801 | // |
| 802 | // UTF-7 encoding table |
| 803 | // |
| 804 | // 0 - Set D (directly encoded characters) |
| 805 | // 1 - Set O (optional direct characters) |
| 806 | // 2 - whitespace characters (optional) |
| 807 | // 3 - special characters |
| 808 | // |
| 809 | static const unsigned char utf7encode[128] = |
| 810 | { |
| 811 | 0, 3, 3, 3, 3, 3, 3, 3, 3, 2, 2, 3, 3, 2, 3, 3, |
| 812 | 3, 3, 3, 3, 3, 3, 3, 3, 3, 3, 3, 3, 3, 3, 3, 3, |
| 813 | 2, 1, 1, 1, 1, 1, 1, 0, 0, 0, 1, 3, 0, 0, 0, 3, |
| 814 | 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 1, 1, 1, 1, 0, |
| 815 | 1, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, |
| 816 | 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 1, 3, 1, 1, 1, |
| 817 | 1, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, |
| 818 | 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 1, 1, 1, 3, 3 |
| 819 | }; |
| 820 | |
| 821 | static inline bool wxIsUTF7Direct(wchar_t wc) |
| 822 | { |
| 823 | return wc < 0x80 && utf7encode[wc] < 1; |
| 824 | } |
| 825 | |
| 826 | size_t wxMBConvUTF7::FromWChar(char *dst, size_t dstLen, |
| 827 | const wchar_t *src, size_t srcLen) const |
| 828 | { |
| 829 | EncoderState stateOrig, |
| 830 | *statePtr; |
| 831 | if ( srcLen == wxNO_LEN ) |
| 832 | { |
| 833 | // we don't apply the stored state when operating on entire strings at |
| 834 | // once |
| 835 | statePtr = &stateOrig; |
| 836 | |
| 837 | srcLen = wxWcslen(src) + 1; |
| 838 | } |
| 839 | else // do use the mode we left the output in previously |
| 840 | { |
| 841 | stateOrig = m_stateEncoder; |
| 842 | statePtr = const_cast<EncoderState *>(&m_stateEncoder); |
| 843 | } |
| 844 | |
| 845 | EncoderState& state = *statePtr; |
| 846 | |
| 847 | |
| 848 | size_t len = 0; |
| 849 | |
| 850 | const wchar_t * const srcEnd = src + srcLen; |
| 851 | while ( src < srcEnd && (!dst || len < dstLen) ) |
| 852 | { |
| 853 | wchar_t cc = *src++; |
| 854 | if ( wxIsUTF7Direct(cc) ) |
| 855 | { |
| 856 | if ( state.IsShifted() ) |
| 857 | { |
| 858 | // pad with zeros the last encoded block if necessary |
| 859 | if ( state.bit ) |
| 860 | { |
| 861 | if ( dst ) |
| 862 | *dst++ = utf7enb64[((state.accum % 16) << (6 - state.bit)) % 64]; |
| 863 | len++; |
| 864 | } |
| 865 | |
| 866 | state.ToDirect(); |
| 867 | |
| 868 | if ( dst ) |
| 869 | *dst++ = '-'; |
| 870 | len++; |
| 871 | } |
| 872 | |
| 873 | if ( dst ) |
| 874 | *dst++ = (char)cc; |
| 875 | len++; |
| 876 | } |
| 877 | else if ( cc == '+' && state.IsDirect() ) |
| 878 | { |
| 879 | if ( dst ) |
| 880 | { |
| 881 | *dst++ = '+'; |
| 882 | *dst++ = '-'; |
| 883 | } |
| 884 | |
| 885 | len += 2; |
| 886 | } |
| 887 | #ifndef WC_UTF16 |
| 888 | else if (((wxUint32)cc) > 0xffff) |
| 889 | { |
| 890 | // no surrogate pair generation (yet?) |
| 891 | return wxCONV_FAILED; |
| 892 | } |
| 893 | #endif |
| 894 | else |
| 895 | { |
| 896 | if ( state.IsDirect() ) |
| 897 | { |
| 898 | state.ToShifted(); |
| 899 | |
| 900 | if ( dst ) |
| 901 | *dst++ = '+'; |
| 902 | len++; |
| 903 | } |
| 904 | |
| 905 | // BASE64 encode string |
| 906 | for ( ;; ) |
| 907 | { |
| 908 | for ( unsigned lsb = 0; lsb < 2; lsb++ ) |
| 909 | { |
| 910 | state.accum <<= 8; |
| 911 | state.accum += lsb ? cc & 0xff : (cc & 0xff00) >> 8; |
| 912 | |
| 913 | for (state.bit += 8; state.bit >= 6; ) |
| 914 | { |
| 915 | state.bit -= 6; |
| 916 | if ( dst ) |
| 917 | *dst++ = utf7enb64[(state.accum >> state.bit) % 64]; |
| 918 | len++; |
| 919 | } |
| 920 | } |
| 921 | |
| 922 | if ( src == srcEnd || wxIsUTF7Direct(cc = *src) ) |
| 923 | break; |
| 924 | |
| 925 | src++; |
| 926 | } |
| 927 | } |
| 928 | } |
| 929 | |
| 930 | // we need to restore the original encoder state if we were called just to |
| 931 | // calculate the amount of space needed as we will presumably be called |
| 932 | // again to really convert the data now |
| 933 | if ( !dst ) |
| 934 | state = stateOrig; |
| 935 | |
| 936 | return len; |
| 937 | } |
| 938 | |
| 939 | // ---------------------------------------------------------------------------- |
| 940 | // UTF-8 |
| 941 | // ---------------------------------------------------------------------------- |
| 942 | |
| 943 | static const wxUint32 utf8_max[]= |
| 944 | { 0x7f, 0x7ff, 0xffff, 0x1fffff, 0x3ffffff, 0x7fffffff, 0xffffffff }; |
| 945 | |
| 946 | // boundaries of the private use area we use to (temporarily) remap invalid |
| 947 | // characters invalid in a UTF-8 encoded string |
| 948 | const wxUint32 wxUnicodePUA = 0x100000; |
| 949 | const wxUint32 wxUnicodePUAEnd = wxUnicodePUA + 256; |
| 950 | |
| 951 | // this table gives the length of the UTF-8 encoding from its first character: |
| 952 | const unsigned char tableUtf8Lengths[256] = { |
| 953 | // single-byte sequences (ASCII): |
| 954 | 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, // 00..0F |
| 955 | 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, // 10..1F |
| 956 | 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, // 20..2F |
| 957 | 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, // 30..3F |
| 958 | 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, // 40..4F |
| 959 | 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, // 50..5F |
| 960 | 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, // 60..6F |
| 961 | 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, // 70..7F |
| 962 | |
| 963 | // these are invalid: |
| 964 | 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, // 80..8F |
| 965 | 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, // 90..9F |
| 966 | 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, // A0..AF |
| 967 | 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, // B0..BF |
| 968 | 0, 0, // C0,C1 |
| 969 | |
| 970 | // two-byte sequences: |
| 971 | 2, 2, 2, 2, 2, 2, 2, 2, 2, 2, 2, 2, 2, 2, // C2..CF |
| 972 | 2, 2, 2, 2, 2, 2, 2, 2, 2, 2, 2, 2, 2, 2, 2, 2, // D0..DF |
| 973 | |
| 974 | // three-byte sequences: |
| 975 | 3, 3, 3, 3, 3, 3, 3, 3, 3, 3, 3, 3, 3, 3, 3, 3, // E0..EF |
| 976 | |
| 977 | // four-byte sequences: |
| 978 | 4, 4, 4, 4, 4, // F0..F4 |
| 979 | |
| 980 | // these are invalid again (5- or 6-byte |
| 981 | // sequences and sequences for code points |
| 982 | // above U+10FFFF, as restricted by RFC 3629): |
| 983 | 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0 // F5..FF |
| 984 | }; |
| 985 | |
| 986 | size_t |
| 987 | wxMBConvStrictUTF8::ToWChar(wchar_t *dst, size_t dstLen, |
| 988 | const char *src, size_t srcLen) const |
| 989 | { |
| 990 | wchar_t *out = dstLen ? dst : NULL; |
| 991 | size_t written = 0; |
| 992 | |
| 993 | if ( srcLen == wxNO_LEN ) |
| 994 | srcLen = strlen(src) + 1; |
| 995 | |
| 996 | for ( const char *p = src; ; p++ ) |
| 997 | { |
| 998 | if ( (srcLen == wxNO_LEN ? !*p : !srcLen) ) |
| 999 | { |
| 1000 | // all done successfully, just add the trailing NULL if we are not |
| 1001 | // using explicit length |
| 1002 | if ( srcLen == wxNO_LEN ) |
| 1003 | { |
| 1004 | if ( out ) |
| 1005 | { |
| 1006 | if ( !dstLen ) |
| 1007 | break; |
| 1008 | |
| 1009 | *out = L'\0'; |
| 1010 | } |
| 1011 | |
| 1012 | written++; |
| 1013 | } |
| 1014 | |
| 1015 | return written; |
| 1016 | } |
| 1017 | |
| 1018 | if ( out && !dstLen-- ) |
| 1019 | break; |
| 1020 | |
| 1021 | wxUint32 code; |
| 1022 | unsigned char c = *p; |
| 1023 | |
| 1024 | if ( c < 0x80 ) |
| 1025 | { |
| 1026 | if ( srcLen == 0 ) // the test works for wxNO_LEN too |
| 1027 | break; |
| 1028 | |
| 1029 | if ( srcLen != wxNO_LEN ) |
| 1030 | srcLen--; |
| 1031 | |
| 1032 | code = c; |
| 1033 | } |
| 1034 | else |
| 1035 | { |
| 1036 | unsigned len = tableUtf8Lengths[c]; |
| 1037 | if ( !len ) |
| 1038 | break; |
| 1039 | |
| 1040 | if ( srcLen < len ) // the test works for wxNO_LEN too |
| 1041 | break; |
| 1042 | |
| 1043 | if ( srcLen != wxNO_LEN ) |
| 1044 | srcLen -= len; |
| 1045 | |
| 1046 | // Char. number range | UTF-8 octet sequence |
| 1047 | // (hexadecimal) | (binary) |
| 1048 | // ----------------------+---------------------------------------- |
| 1049 | // 0000 0000 - 0000 007F | 0xxxxxxx |
| 1050 | // 0000 0080 - 0000 07FF | 110xxxxx 10xxxxxx |
| 1051 | // 0000 0800 - 0000 FFFF | 1110xxxx 10xxxxxx 10xxxxxx |
| 1052 | // 0001 0000 - 0010 FFFF | 11110xxx 10xxxxxx 10xxxxxx 10xxxxxx |
| 1053 | // |
| 1054 | // Code point value is stored in bits marked with 'x', |
| 1055 | // lowest-order bit of the value on the right side in the diagram |
| 1056 | // above. (from RFC 3629) |
| 1057 | |
| 1058 | // mask to extract lead byte's value ('x' bits above), by sequence |
| 1059 | // length: |
| 1060 | static const unsigned char leadValueMask[] = { 0x7F, 0x1F, 0x0F, 0x07 }; |
| 1061 | |
| 1062 | // mask and value of lead byte's most significant bits, by length: |
| 1063 | static const unsigned char leadMarkerMask[] = { 0x80, 0xE0, 0xF0, 0xF8 }; |
| 1064 | static const unsigned char leadMarkerVal[] = { 0x00, 0xC0, 0xE0, 0xF0 }; |
| 1065 | |
| 1066 | len--; // it's more convenient to work with 0-based length here |
| 1067 | |
| 1068 | // extract the lead byte's value bits: |
| 1069 | if ( (c & leadMarkerMask[len]) != leadMarkerVal[len] ) |
| 1070 | break; |
| 1071 | |
| 1072 | code = c & leadValueMask[len]; |
| 1073 | |
| 1074 | // all remaining bytes, if any, are handled in the same way |
| 1075 | // regardless of sequence's length: |
| 1076 | for ( ; len; --len ) |
| 1077 | { |
| 1078 | c = *++p; |
| 1079 | if ( (c & 0xC0) != 0x80 ) |
| 1080 | return wxCONV_FAILED; |
| 1081 | |
| 1082 | code <<= 6; |
| 1083 | code |= c & 0x3F; |
| 1084 | } |
| 1085 | } |
| 1086 | |
| 1087 | #ifdef WC_UTF16 |
| 1088 | // cast is ok because wchar_t == wxUint16 if WC_UTF16 |
| 1089 | if ( encode_utf16(code, (wxUint16 *)out) == 2 ) |
| 1090 | { |
| 1091 | if ( out ) |
| 1092 | out++; |
| 1093 | written++; |
| 1094 | } |
| 1095 | #else // !WC_UTF16 |
| 1096 | if ( out ) |
| 1097 | *out = code; |
| 1098 | #endif // WC_UTF16/!WC_UTF16 |
| 1099 | |
| 1100 | if ( out ) |
| 1101 | out++; |
| 1102 | |
| 1103 | written++; |
| 1104 | } |
| 1105 | |
| 1106 | return wxCONV_FAILED; |
| 1107 | } |
| 1108 | |
| 1109 | size_t |
| 1110 | wxMBConvStrictUTF8::FromWChar(char *dst, size_t dstLen, |
| 1111 | const wchar_t *src, size_t srcLen) const |
| 1112 | { |
| 1113 | char *out = dstLen ? dst : NULL; |
| 1114 | size_t written = 0; |
| 1115 | |
| 1116 | for ( const wchar_t *wp = src; ; wp++ ) |
| 1117 | { |
| 1118 | if ( (srcLen == wxNO_LEN ? !*wp : !srcLen) ) |
| 1119 | { |
| 1120 | // all done successfully, just add the trailing NULL if we are not |
| 1121 | // using explicit length |
| 1122 | if ( srcLen == wxNO_LEN ) |
| 1123 | { |
| 1124 | if ( out ) |
| 1125 | { |
| 1126 | if ( !dstLen ) |
| 1127 | break; |
| 1128 | |
| 1129 | *out = '\0'; |
| 1130 | } |
| 1131 | |
| 1132 | written++; |
| 1133 | } |
| 1134 | |
| 1135 | return written; |
| 1136 | } |
| 1137 | |
| 1138 | if ( srcLen != wxNO_LEN ) |
| 1139 | srcLen--; |
| 1140 | |
| 1141 | wxUint32 code; |
| 1142 | #ifdef WC_UTF16 |
| 1143 | // cast is ok for WC_UTF16 |
| 1144 | if ( decode_utf16((const wxUint16 *)wp, code) == 2 ) |
| 1145 | { |
| 1146 | // skip the next char too as we decoded a surrogate |
| 1147 | wp++; |
| 1148 | } |
| 1149 | #else // wchar_t is UTF-32 |
| 1150 | code = *wp & 0x7fffffff; |
| 1151 | #endif |
| 1152 | |
| 1153 | unsigned len; |
| 1154 | if ( code <= 0x7F ) |
| 1155 | { |
| 1156 | len = 1; |
| 1157 | if ( out ) |
| 1158 | { |
| 1159 | if ( dstLen < len ) |
| 1160 | break; |
| 1161 | |
| 1162 | out[0] = (char)code; |
| 1163 | } |
| 1164 | } |
| 1165 | else if ( code <= 0x07FF ) |
| 1166 | { |
| 1167 | len = 2; |
| 1168 | if ( out ) |
| 1169 | { |
| 1170 | if ( dstLen < len ) |
| 1171 | break; |
| 1172 | |
| 1173 | // NB: this line takes 6 least significant bits, encodes them as |
| 1174 | // 10xxxxxx and discards them so that the next byte can be encoded: |
| 1175 | out[1] = 0x80 | (code & 0x3F); code >>= 6; |
| 1176 | out[0] = 0xC0 | code; |
| 1177 | } |
| 1178 | } |
| 1179 | else if ( code < 0xFFFF ) |
| 1180 | { |
| 1181 | len = 3; |
| 1182 | if ( out ) |
| 1183 | { |
| 1184 | if ( dstLen < len ) |
| 1185 | break; |
| 1186 | |
| 1187 | out[2] = 0x80 | (code & 0x3F); code >>= 6; |
| 1188 | out[1] = 0x80 | (code & 0x3F); code >>= 6; |
| 1189 | out[0] = 0xE0 | code; |
| 1190 | } |
| 1191 | } |
| 1192 | else if ( code <= 0x10FFFF ) |
| 1193 | { |
| 1194 | len = 4; |
| 1195 | if ( out ) |
| 1196 | { |
| 1197 | if ( dstLen < len ) |
| 1198 | break; |
| 1199 | |
| 1200 | out[3] = 0x80 | (code & 0x3F); code >>= 6; |
| 1201 | out[2] = 0x80 | (code & 0x3F); code >>= 6; |
| 1202 | out[1] = 0x80 | (code & 0x3F); code >>= 6; |
| 1203 | out[0] = 0xF0 | code; |
| 1204 | } |
| 1205 | } |
| 1206 | else |
| 1207 | { |
| 1208 | wxFAIL_MSG( wxT("trying to encode undefined Unicode character") ); |
| 1209 | break; |
| 1210 | } |
| 1211 | |
| 1212 | if ( out ) |
| 1213 | { |
| 1214 | out += len; |
| 1215 | dstLen -= len; |
| 1216 | } |
| 1217 | |
| 1218 | written += len; |
| 1219 | } |
| 1220 | |
| 1221 | // we only get here if an error occurs during decoding |
| 1222 | return wxCONV_FAILED; |
| 1223 | } |
| 1224 | |
| 1225 | size_t wxMBConvUTF8::ToWChar(wchar_t *buf, size_t n, |
| 1226 | const char *psz, size_t srcLen) const |
| 1227 | { |
| 1228 | if ( m_options == MAP_INVALID_UTF8_NOT ) |
| 1229 | return wxMBConvStrictUTF8::ToWChar(buf, n, psz, srcLen); |
| 1230 | |
| 1231 | size_t len = 0; |
| 1232 | |
| 1233 | while ((srcLen == wxNO_LEN ? *psz : srcLen--) && ((!buf) || (len < n))) |
| 1234 | { |
| 1235 | const char *opsz = psz; |
| 1236 | bool invalid = false; |
| 1237 | unsigned char cc = *psz++, fc = cc; |
| 1238 | unsigned cnt; |
| 1239 | for (cnt = 0; fc & 0x80; cnt++) |
| 1240 | fc <<= 1; |
| 1241 | |
| 1242 | if (!cnt) |
| 1243 | { |
| 1244 | // plain ASCII char |
| 1245 | if (buf) |
| 1246 | *buf++ = cc; |
| 1247 | len++; |
| 1248 | |
| 1249 | // escape the escape character for octal escapes |
| 1250 | if ((m_options & MAP_INVALID_UTF8_TO_OCTAL) |
| 1251 | && cc == '\\' && (!buf || len < n)) |
| 1252 | { |
| 1253 | if (buf) |
| 1254 | *buf++ = cc; |
| 1255 | len++; |
| 1256 | } |
| 1257 | } |
| 1258 | else |
| 1259 | { |
| 1260 | cnt--; |
| 1261 | if (!cnt) |
| 1262 | { |
| 1263 | // invalid UTF-8 sequence |
| 1264 | invalid = true; |
| 1265 | } |
| 1266 | else |
| 1267 | { |
| 1268 | unsigned ocnt = cnt - 1; |
| 1269 | wxUint32 res = cc & (0x3f >> cnt); |
| 1270 | while (cnt--) |
| 1271 | { |
| 1272 | cc = *psz; |
| 1273 | if ((cc & 0xC0) != 0x80) |
| 1274 | { |
| 1275 | // invalid UTF-8 sequence |
| 1276 | invalid = true; |
| 1277 | break; |
| 1278 | } |
| 1279 | |
| 1280 | psz++; |
| 1281 | res = (res << 6) | (cc & 0x3f); |
| 1282 | } |
| 1283 | |
| 1284 | if (invalid || res <= utf8_max[ocnt]) |
| 1285 | { |
| 1286 | // illegal UTF-8 encoding |
| 1287 | invalid = true; |
| 1288 | } |
| 1289 | else if ((m_options & MAP_INVALID_UTF8_TO_PUA) && |
| 1290 | res >= wxUnicodePUA && res < wxUnicodePUAEnd) |
| 1291 | { |
| 1292 | // if one of our PUA characters turns up externally |
| 1293 | // it must also be treated as an illegal sequence |
| 1294 | // (a bit like you have to escape an escape character) |
| 1295 | invalid = true; |
| 1296 | } |
| 1297 | else |
| 1298 | { |
| 1299 | #ifdef WC_UTF16 |
| 1300 | // cast is ok because wchar_t == wxUint16 if WC_UTF16 |
| 1301 | size_t pa = encode_utf16(res, (wxUint16 *)buf); |
| 1302 | if (pa == wxCONV_FAILED) |
| 1303 | { |
| 1304 | invalid = true; |
| 1305 | } |
| 1306 | else |
| 1307 | { |
| 1308 | if (buf) |
| 1309 | buf += pa; |
| 1310 | len += pa; |
| 1311 | } |
| 1312 | #else // !WC_UTF16 |
| 1313 | if (buf) |
| 1314 | *buf++ = (wchar_t)res; |
| 1315 | len++; |
| 1316 | #endif // WC_UTF16/!WC_UTF16 |
| 1317 | } |
| 1318 | } |
| 1319 | |
| 1320 | if (invalid) |
| 1321 | { |
| 1322 | if (m_options & MAP_INVALID_UTF8_TO_PUA) |
| 1323 | { |
| 1324 | while (opsz < psz && (!buf || len < n)) |
| 1325 | { |
| 1326 | #ifdef WC_UTF16 |
| 1327 | // cast is ok because wchar_t == wxUuint16 if WC_UTF16 |
| 1328 | size_t pa = encode_utf16((unsigned char)*opsz + wxUnicodePUA, (wxUint16 *)buf); |
| 1329 | wxASSERT(pa != wxCONV_FAILED); |
| 1330 | if (buf) |
| 1331 | buf += pa; |
| 1332 | opsz++; |
| 1333 | len += pa; |
| 1334 | #else |
| 1335 | if (buf) |
| 1336 | *buf++ = (wchar_t)(wxUnicodePUA + (unsigned char)*opsz); |
| 1337 | opsz++; |
| 1338 | len++; |
| 1339 | #endif |
| 1340 | } |
| 1341 | } |
| 1342 | else if (m_options & MAP_INVALID_UTF8_TO_OCTAL) |
| 1343 | { |
| 1344 | while (opsz < psz && (!buf || len < n)) |
| 1345 | { |
| 1346 | if ( buf && len + 3 < n ) |
| 1347 | { |
| 1348 | unsigned char on = *opsz; |
| 1349 | *buf++ = L'\\'; |
| 1350 | *buf++ = (wchar_t)( L'0' + on / 0100 ); |
| 1351 | *buf++ = (wchar_t)( L'0' + (on % 0100) / 010 ); |
| 1352 | *buf++ = (wchar_t)( L'0' + on % 010 ); |
| 1353 | } |
| 1354 | |
| 1355 | opsz++; |
| 1356 | len += 4; |
| 1357 | } |
| 1358 | } |
| 1359 | else // MAP_INVALID_UTF8_NOT |
| 1360 | { |
| 1361 | return wxCONV_FAILED; |
| 1362 | } |
| 1363 | } |
| 1364 | } |
| 1365 | } |
| 1366 | |
| 1367 | if (srcLen == wxNO_LEN && buf && (len < n)) |
| 1368 | *buf = 0; |
| 1369 | |
| 1370 | return len + 1; |
| 1371 | } |
| 1372 | |
| 1373 | static inline bool isoctal(wchar_t wch) |
| 1374 | { |
| 1375 | return L'0' <= wch && wch <= L'7'; |
| 1376 | } |
| 1377 | |
| 1378 | size_t wxMBConvUTF8::FromWChar(char *buf, size_t n, |
| 1379 | const wchar_t *psz, size_t srcLen) const |
| 1380 | { |
| 1381 | if ( m_options == MAP_INVALID_UTF8_NOT ) |
| 1382 | return wxMBConvStrictUTF8::FromWChar(buf, n, psz, srcLen); |
| 1383 | |
| 1384 | size_t len = 0; |
| 1385 | |
| 1386 | while ((srcLen == wxNO_LEN ? *psz : srcLen--) && ((!buf) || (len < n))) |
| 1387 | { |
| 1388 | wxUint32 cc; |
| 1389 | |
| 1390 | #ifdef WC_UTF16 |
| 1391 | // cast is ok for WC_UTF16 |
| 1392 | size_t pa = decode_utf16((const wxUint16 *)psz, cc); |
| 1393 | psz += (pa == wxCONV_FAILED) ? 1 : pa; |
| 1394 | #else |
| 1395 | cc = (*psz++) & 0x7fffffff; |
| 1396 | #endif |
| 1397 | |
| 1398 | if ( (m_options & MAP_INVALID_UTF8_TO_PUA) |
| 1399 | && cc >= wxUnicodePUA && cc < wxUnicodePUAEnd ) |
| 1400 | { |
| 1401 | if (buf) |
| 1402 | *buf++ = (char)(cc - wxUnicodePUA); |
| 1403 | len++; |
| 1404 | } |
| 1405 | else if ( (m_options & MAP_INVALID_UTF8_TO_OCTAL) |
| 1406 | && cc == L'\\' && psz[0] == L'\\' ) |
| 1407 | { |
| 1408 | if (buf) |
| 1409 | *buf++ = (char)cc; |
| 1410 | psz++; |
| 1411 | len++; |
| 1412 | } |
| 1413 | else if ( (m_options & MAP_INVALID_UTF8_TO_OCTAL) && |
| 1414 | cc == L'\\' && |
| 1415 | isoctal(psz[0]) && isoctal(psz[1]) && isoctal(psz[2]) ) |
| 1416 | { |
| 1417 | if (buf) |
| 1418 | { |
| 1419 | *buf++ = (char) ((psz[0] - L'0') * 0100 + |
| 1420 | (psz[1] - L'0') * 010 + |
| 1421 | (psz[2] - L'0')); |
| 1422 | } |
| 1423 | |
| 1424 | psz += 3; |
| 1425 | len++; |
| 1426 | } |
| 1427 | else |
| 1428 | { |
| 1429 | unsigned cnt; |
| 1430 | for (cnt = 0; cc > utf8_max[cnt]; cnt++) |
| 1431 | { |
| 1432 | } |
| 1433 | |
| 1434 | if (!cnt) |
| 1435 | { |
| 1436 | // plain ASCII char |
| 1437 | if (buf) |
| 1438 | *buf++ = (char) cc; |
| 1439 | len++; |
| 1440 | } |
| 1441 | else |
| 1442 | { |
| 1443 | len += cnt + 1; |
| 1444 | if (buf) |
| 1445 | { |
| 1446 | *buf++ = (char) ((-128 >> cnt) | ((cc >> (cnt * 6)) & (0x3f >> cnt))); |
| 1447 | while (cnt--) |
| 1448 | *buf++ = (char) (0x80 | ((cc >> (cnt * 6)) & 0x3f)); |
| 1449 | } |
| 1450 | } |
| 1451 | } |
| 1452 | } |
| 1453 | |
| 1454 | if (srcLen == wxNO_LEN && buf && (len < n)) |
| 1455 | *buf = 0; |
| 1456 | |
| 1457 | return len + 1; |
| 1458 | } |
| 1459 | |
| 1460 | // ============================================================================ |
| 1461 | // UTF-16 |
| 1462 | // ============================================================================ |
| 1463 | |
| 1464 | #ifdef WORDS_BIGENDIAN |
| 1465 | #define wxMBConvUTF16straight wxMBConvUTF16BE |
| 1466 | #define wxMBConvUTF16swap wxMBConvUTF16LE |
| 1467 | #else |
| 1468 | #define wxMBConvUTF16swap wxMBConvUTF16BE |
| 1469 | #define wxMBConvUTF16straight wxMBConvUTF16LE |
| 1470 | #endif |
| 1471 | |
| 1472 | /* static */ |
| 1473 | size_t wxMBConvUTF16Base::GetLength(const char *src, size_t srcLen) |
| 1474 | { |
| 1475 | if ( srcLen == wxNO_LEN ) |
| 1476 | { |
| 1477 | // count the number of bytes in input, including the trailing NULs |
| 1478 | const wxUint16 *inBuff = reinterpret_cast<const wxUint16 *>(src); |
| 1479 | for ( srcLen = 1; *inBuff++; srcLen++ ) |
| 1480 | ; |
| 1481 | |
| 1482 | srcLen *= BYTES_PER_CHAR; |
| 1483 | } |
| 1484 | else // we already have the length |
| 1485 | { |
| 1486 | // we can only convert an entire number of UTF-16 characters |
| 1487 | if ( srcLen % BYTES_PER_CHAR ) |
| 1488 | return wxCONV_FAILED; |
| 1489 | } |
| 1490 | |
| 1491 | return srcLen; |
| 1492 | } |
| 1493 | |
| 1494 | // case when in-memory representation is UTF-16 too |
| 1495 | #ifdef WC_UTF16 |
| 1496 | |
| 1497 | // ---------------------------------------------------------------------------- |
| 1498 | // conversions without endianness change |
| 1499 | // ---------------------------------------------------------------------------- |
| 1500 | |
| 1501 | size_t |
| 1502 | wxMBConvUTF16straight::ToWChar(wchar_t *dst, size_t dstLen, |
| 1503 | const char *src, size_t srcLen) const |
| 1504 | { |
| 1505 | // set up the scene for using memcpy() (which is presumably more efficient |
| 1506 | // than copying the bytes one by one) |
| 1507 | srcLen = GetLength(src, srcLen); |
| 1508 | if ( srcLen == wxNO_LEN ) |
| 1509 | return wxCONV_FAILED; |
| 1510 | |
| 1511 | const size_t inLen = srcLen / BYTES_PER_CHAR; |
| 1512 | if ( dst ) |
| 1513 | { |
| 1514 | if ( dstLen < inLen ) |
| 1515 | return wxCONV_FAILED; |
| 1516 | |
| 1517 | memcpy(dst, src, srcLen); |
| 1518 | } |
| 1519 | |
| 1520 | return inLen; |
| 1521 | } |
| 1522 | |
| 1523 | size_t |
| 1524 | wxMBConvUTF16straight::FromWChar(char *dst, size_t dstLen, |
| 1525 | const wchar_t *src, size_t srcLen) const |
| 1526 | { |
| 1527 | if ( srcLen == wxNO_LEN ) |
| 1528 | srcLen = wxWcslen(src) + 1; |
| 1529 | |
| 1530 | srcLen *= BYTES_PER_CHAR; |
| 1531 | |
| 1532 | if ( dst ) |
| 1533 | { |
| 1534 | if ( dstLen < srcLen ) |
| 1535 | return wxCONV_FAILED; |
| 1536 | |
| 1537 | memcpy(dst, src, srcLen); |
| 1538 | } |
| 1539 | |
| 1540 | return srcLen; |
| 1541 | } |
| 1542 | |
| 1543 | // ---------------------------------------------------------------------------- |
| 1544 | // endian-reversing conversions |
| 1545 | // ---------------------------------------------------------------------------- |
| 1546 | |
| 1547 | size_t |
| 1548 | wxMBConvUTF16swap::ToWChar(wchar_t *dst, size_t dstLen, |
| 1549 | const char *src, size_t srcLen) const |
| 1550 | { |
| 1551 | srcLen = GetLength(src, srcLen); |
| 1552 | if ( srcLen == wxNO_LEN ) |
| 1553 | return wxCONV_FAILED; |
| 1554 | |
| 1555 | srcLen /= BYTES_PER_CHAR; |
| 1556 | |
| 1557 | if ( dst ) |
| 1558 | { |
| 1559 | if ( dstLen < srcLen ) |
| 1560 | return wxCONV_FAILED; |
| 1561 | |
| 1562 | const wxUint16 *inBuff = reinterpret_cast<const wxUint16 *>(src); |
| 1563 | for ( size_t n = 0; n < srcLen; n++, inBuff++ ) |
| 1564 | { |
| 1565 | *dst++ = wxUINT16_SWAP_ALWAYS(*inBuff); |
| 1566 | } |
| 1567 | } |
| 1568 | |
| 1569 | return srcLen; |
| 1570 | } |
| 1571 | |
| 1572 | size_t |
| 1573 | wxMBConvUTF16swap::FromWChar(char *dst, size_t dstLen, |
| 1574 | const wchar_t *src, size_t srcLen) const |
| 1575 | { |
| 1576 | if ( srcLen == wxNO_LEN ) |
| 1577 | srcLen = wxWcslen(src) + 1; |
| 1578 | |
| 1579 | srcLen *= BYTES_PER_CHAR; |
| 1580 | |
| 1581 | if ( dst ) |
| 1582 | { |
| 1583 | if ( dstLen < srcLen ) |
| 1584 | return wxCONV_FAILED; |
| 1585 | |
| 1586 | wxUint16 *outBuff = reinterpret_cast<wxUint16 *>(dst); |
| 1587 | for ( size_t n = 0; n < srcLen; n += BYTES_PER_CHAR, src++ ) |
| 1588 | { |
| 1589 | *outBuff++ = wxUINT16_SWAP_ALWAYS(*src); |
| 1590 | } |
| 1591 | } |
| 1592 | |
| 1593 | return srcLen; |
| 1594 | } |
| 1595 | |
| 1596 | #else // !WC_UTF16: wchar_t is UTF-32 |
| 1597 | |
| 1598 | // ---------------------------------------------------------------------------- |
| 1599 | // conversions without endianness change |
| 1600 | // ---------------------------------------------------------------------------- |
| 1601 | |
| 1602 | size_t |
| 1603 | wxMBConvUTF16straight::ToWChar(wchar_t *dst, size_t dstLen, |
| 1604 | const char *src, size_t srcLen) const |
| 1605 | { |
| 1606 | srcLen = GetLength(src, srcLen); |
| 1607 | if ( srcLen == wxNO_LEN ) |
| 1608 | return wxCONV_FAILED; |
| 1609 | |
| 1610 | const size_t inLen = srcLen / BYTES_PER_CHAR; |
| 1611 | if ( !dst ) |
| 1612 | { |
| 1613 | // optimization: return maximal space which could be needed for this |
| 1614 | // string even if the real size could be smaller if the buffer contains |
| 1615 | // any surrogates |
| 1616 | return inLen; |
| 1617 | } |
| 1618 | |
| 1619 | size_t outLen = 0; |
| 1620 | const wxUint16 *inBuff = reinterpret_cast<const wxUint16 *>(src); |
| 1621 | for ( const wxUint16 * const inEnd = inBuff + inLen; inBuff < inEnd; ) |
| 1622 | { |
| 1623 | const wxUint32 ch = wxDecodeSurrogate(&inBuff); |
| 1624 | if ( !inBuff ) |
| 1625 | return wxCONV_FAILED; |
| 1626 | |
| 1627 | if ( ++outLen > dstLen ) |
| 1628 | return wxCONV_FAILED; |
| 1629 | |
| 1630 | *dst++ = ch; |
| 1631 | } |
| 1632 | |
| 1633 | |
| 1634 | return outLen; |
| 1635 | } |
| 1636 | |
| 1637 | size_t |
| 1638 | wxMBConvUTF16straight::FromWChar(char *dst, size_t dstLen, |
| 1639 | const wchar_t *src, size_t srcLen) const |
| 1640 | { |
| 1641 | if ( srcLen == wxNO_LEN ) |
| 1642 | srcLen = wxWcslen(src) + 1; |
| 1643 | |
| 1644 | size_t outLen = 0; |
| 1645 | wxUint16 *outBuff = reinterpret_cast<wxUint16 *>(dst); |
| 1646 | for ( size_t n = 0; n < srcLen; n++ ) |
| 1647 | { |
| 1648 | wxUint16 cc[2] = { 0 }; |
| 1649 | const size_t numChars = encode_utf16(*src++, cc); |
| 1650 | if ( numChars == wxCONV_FAILED ) |
| 1651 | return wxCONV_FAILED; |
| 1652 | |
| 1653 | outLen += numChars * BYTES_PER_CHAR; |
| 1654 | if ( outBuff ) |
| 1655 | { |
| 1656 | if ( outLen > dstLen ) |
| 1657 | return wxCONV_FAILED; |
| 1658 | |
| 1659 | *outBuff++ = cc[0]; |
| 1660 | if ( numChars == 2 ) |
| 1661 | { |
| 1662 | // second character of a surrogate |
| 1663 | *outBuff++ = cc[1]; |
| 1664 | } |
| 1665 | } |
| 1666 | } |
| 1667 | |
| 1668 | return outLen; |
| 1669 | } |
| 1670 | |
| 1671 | // ---------------------------------------------------------------------------- |
| 1672 | // endian-reversing conversions |
| 1673 | // ---------------------------------------------------------------------------- |
| 1674 | |
| 1675 | size_t |
| 1676 | wxMBConvUTF16swap::ToWChar(wchar_t *dst, size_t dstLen, |
| 1677 | const char *src, size_t srcLen) const |
| 1678 | { |
| 1679 | srcLen = GetLength(src, srcLen); |
| 1680 | if ( srcLen == wxNO_LEN ) |
| 1681 | return wxCONV_FAILED; |
| 1682 | |
| 1683 | const size_t inLen = srcLen / BYTES_PER_CHAR; |
| 1684 | if ( !dst ) |
| 1685 | { |
| 1686 | // optimization: return maximal space which could be needed for this |
| 1687 | // string even if the real size could be smaller if the buffer contains |
| 1688 | // any surrogates |
| 1689 | return inLen; |
| 1690 | } |
| 1691 | |
| 1692 | size_t outLen = 0; |
| 1693 | const wxUint16 *inBuff = reinterpret_cast<const wxUint16 *>(src); |
| 1694 | for ( const wxUint16 * const inEnd = inBuff + inLen; inBuff < inEnd; ) |
| 1695 | { |
| 1696 | wxUint32 ch; |
| 1697 | wxUint16 tmp[2]; |
| 1698 | |
| 1699 | tmp[0] = wxUINT16_SWAP_ALWAYS(*inBuff); |
| 1700 | inBuff++; |
| 1701 | tmp[1] = wxUINT16_SWAP_ALWAYS(*inBuff); |
| 1702 | |
| 1703 | const size_t numChars = decode_utf16(tmp, ch); |
| 1704 | if ( numChars == wxCONV_FAILED ) |
| 1705 | return wxCONV_FAILED; |
| 1706 | |
| 1707 | if ( numChars == 2 ) |
| 1708 | inBuff++; |
| 1709 | |
| 1710 | if ( ++outLen > dstLen ) |
| 1711 | return wxCONV_FAILED; |
| 1712 | |
| 1713 | *dst++ = ch; |
| 1714 | } |
| 1715 | |
| 1716 | |
| 1717 | return outLen; |
| 1718 | } |
| 1719 | |
| 1720 | size_t |
| 1721 | wxMBConvUTF16swap::FromWChar(char *dst, size_t dstLen, |
| 1722 | const wchar_t *src, size_t srcLen) const |
| 1723 | { |
| 1724 | if ( srcLen == wxNO_LEN ) |
| 1725 | srcLen = wxWcslen(src) + 1; |
| 1726 | |
| 1727 | size_t outLen = 0; |
| 1728 | wxUint16 *outBuff = reinterpret_cast<wxUint16 *>(dst); |
| 1729 | for ( const wchar_t *srcEnd = src + srcLen; src < srcEnd; src++ ) |
| 1730 | { |
| 1731 | wxUint16 cc[2] = { 0 }; |
| 1732 | const size_t numChars = encode_utf16(*src, cc); |
| 1733 | if ( numChars == wxCONV_FAILED ) |
| 1734 | return wxCONV_FAILED; |
| 1735 | |
| 1736 | outLen += numChars * BYTES_PER_CHAR; |
| 1737 | if ( outBuff ) |
| 1738 | { |
| 1739 | if ( outLen > dstLen ) |
| 1740 | return wxCONV_FAILED; |
| 1741 | |
| 1742 | *outBuff++ = wxUINT16_SWAP_ALWAYS(cc[0]); |
| 1743 | if ( numChars == 2 ) |
| 1744 | { |
| 1745 | // second character of a surrogate |
| 1746 | *outBuff++ = wxUINT16_SWAP_ALWAYS(cc[1]); |
| 1747 | } |
| 1748 | } |
| 1749 | } |
| 1750 | |
| 1751 | return outLen; |
| 1752 | } |
| 1753 | |
| 1754 | #endif // WC_UTF16/!WC_UTF16 |
| 1755 | |
| 1756 | |
| 1757 | // ============================================================================ |
| 1758 | // UTF-32 |
| 1759 | // ============================================================================ |
| 1760 | |
| 1761 | #ifdef WORDS_BIGENDIAN |
| 1762 | #define wxMBConvUTF32straight wxMBConvUTF32BE |
| 1763 | #define wxMBConvUTF32swap wxMBConvUTF32LE |
| 1764 | #else |
| 1765 | #define wxMBConvUTF32swap wxMBConvUTF32BE |
| 1766 | #define wxMBConvUTF32straight wxMBConvUTF32LE |
| 1767 | #endif |
| 1768 | |
| 1769 | |
| 1770 | WXDLLIMPEXP_DATA_BASE(wxMBConvUTF32LE) wxConvUTF32LE; |
| 1771 | WXDLLIMPEXP_DATA_BASE(wxMBConvUTF32BE) wxConvUTF32BE; |
| 1772 | |
| 1773 | /* static */ |
| 1774 | size_t wxMBConvUTF32Base::GetLength(const char *src, size_t srcLen) |
| 1775 | { |
| 1776 | if ( srcLen == wxNO_LEN ) |
| 1777 | { |
| 1778 | // count the number of bytes in input, including the trailing NULs |
| 1779 | const wxUint32 *inBuff = reinterpret_cast<const wxUint32 *>(src); |
| 1780 | for ( srcLen = 1; *inBuff++; srcLen++ ) |
| 1781 | ; |
| 1782 | |
| 1783 | srcLen *= BYTES_PER_CHAR; |
| 1784 | } |
| 1785 | else // we already have the length |
| 1786 | { |
| 1787 | // we can only convert an entire number of UTF-32 characters |
| 1788 | if ( srcLen % BYTES_PER_CHAR ) |
| 1789 | return wxCONV_FAILED; |
| 1790 | } |
| 1791 | |
| 1792 | return srcLen; |
| 1793 | } |
| 1794 | |
| 1795 | // case when in-memory representation is UTF-16 |
| 1796 | #ifdef WC_UTF16 |
| 1797 | |
| 1798 | // ---------------------------------------------------------------------------- |
| 1799 | // conversions without endianness change |
| 1800 | // ---------------------------------------------------------------------------- |
| 1801 | |
| 1802 | size_t |
| 1803 | wxMBConvUTF32straight::ToWChar(wchar_t *dst, size_t dstLen, |
| 1804 | const char *src, size_t srcLen) const |
| 1805 | { |
| 1806 | srcLen = GetLength(src, srcLen); |
| 1807 | if ( srcLen == wxNO_LEN ) |
| 1808 | return wxCONV_FAILED; |
| 1809 | |
| 1810 | const wxUint32 *inBuff = reinterpret_cast<const wxUint32 *>(src); |
| 1811 | const size_t inLen = srcLen / BYTES_PER_CHAR; |
| 1812 | size_t outLen = 0; |
| 1813 | for ( size_t n = 0; n < inLen; n++ ) |
| 1814 | { |
| 1815 | wxUint16 cc[2] = { 0 }; |
| 1816 | const size_t numChars = encode_utf16(*inBuff++, cc); |
| 1817 | if ( numChars == wxCONV_FAILED ) |
| 1818 | return wxCONV_FAILED; |
| 1819 | |
| 1820 | outLen += numChars; |
| 1821 | if ( dst ) |
| 1822 | { |
| 1823 | if ( outLen > dstLen ) |
| 1824 | return wxCONV_FAILED; |
| 1825 | |
| 1826 | *dst++ = cc[0]; |
| 1827 | if ( numChars == 2 ) |
| 1828 | { |
| 1829 | // second character of a surrogate |
| 1830 | *dst++ = cc[1]; |
| 1831 | } |
| 1832 | } |
| 1833 | } |
| 1834 | |
| 1835 | return outLen; |
| 1836 | } |
| 1837 | |
| 1838 | size_t |
| 1839 | wxMBConvUTF32straight::FromWChar(char *dst, size_t dstLen, |
| 1840 | const wchar_t *src, size_t srcLen) const |
| 1841 | { |
| 1842 | if ( srcLen == wxNO_LEN ) |
| 1843 | srcLen = wxWcslen(src) + 1; |
| 1844 | |
| 1845 | if ( !dst ) |
| 1846 | { |
| 1847 | // optimization: return maximal space which could be needed for this |
| 1848 | // string instead of the exact amount which could be less if there are |
| 1849 | // any surrogates in the input |
| 1850 | // |
| 1851 | // we consider that surrogates are rare enough to make it worthwhile to |
| 1852 | // avoid running the loop below at the cost of slightly extra memory |
| 1853 | // consumption |
| 1854 | return srcLen * BYTES_PER_CHAR; |
| 1855 | } |
| 1856 | |
| 1857 | wxUint32 *outBuff = reinterpret_cast<wxUint32 *>(dst); |
| 1858 | size_t outLen = 0; |
| 1859 | for ( const wchar_t * const srcEnd = src + srcLen; src < srcEnd; ) |
| 1860 | { |
| 1861 | const wxUint32 ch = wxDecodeSurrogate(&src); |
| 1862 | if ( !src ) |
| 1863 | return wxCONV_FAILED; |
| 1864 | |
| 1865 | outLen += BYTES_PER_CHAR; |
| 1866 | |
| 1867 | if ( outLen > dstLen ) |
| 1868 | return wxCONV_FAILED; |
| 1869 | |
| 1870 | *outBuff++ = ch; |
| 1871 | } |
| 1872 | |
| 1873 | return outLen; |
| 1874 | } |
| 1875 | |
| 1876 | // ---------------------------------------------------------------------------- |
| 1877 | // endian-reversing conversions |
| 1878 | // ---------------------------------------------------------------------------- |
| 1879 | |
| 1880 | size_t |
| 1881 | wxMBConvUTF32swap::ToWChar(wchar_t *dst, size_t dstLen, |
| 1882 | const char *src, size_t srcLen) const |
| 1883 | { |
| 1884 | srcLen = GetLength(src, srcLen); |
| 1885 | if ( srcLen == wxNO_LEN ) |
| 1886 | return wxCONV_FAILED; |
| 1887 | |
| 1888 | const wxUint32 *inBuff = reinterpret_cast<const wxUint32 *>(src); |
| 1889 | const size_t inLen = srcLen / BYTES_PER_CHAR; |
| 1890 | size_t outLen = 0; |
| 1891 | for ( size_t n = 0; n < inLen; n++, inBuff++ ) |
| 1892 | { |
| 1893 | wxUint16 cc[2] = { 0 }; |
| 1894 | const size_t numChars = encode_utf16(wxUINT32_SWAP_ALWAYS(*inBuff), cc); |
| 1895 | if ( numChars == wxCONV_FAILED ) |
| 1896 | return wxCONV_FAILED; |
| 1897 | |
| 1898 | outLen += numChars; |
| 1899 | if ( dst ) |
| 1900 | { |
| 1901 | if ( outLen > dstLen ) |
| 1902 | return wxCONV_FAILED; |
| 1903 | |
| 1904 | *dst++ = cc[0]; |
| 1905 | if ( numChars == 2 ) |
| 1906 | { |
| 1907 | // second character of a surrogate |
| 1908 | *dst++ = cc[1]; |
| 1909 | } |
| 1910 | } |
| 1911 | } |
| 1912 | |
| 1913 | return outLen; |
| 1914 | } |
| 1915 | |
| 1916 | size_t |
| 1917 | wxMBConvUTF32swap::FromWChar(char *dst, size_t dstLen, |
| 1918 | const wchar_t *src, size_t srcLen) const |
| 1919 | { |
| 1920 | if ( srcLen == wxNO_LEN ) |
| 1921 | srcLen = wxWcslen(src) + 1; |
| 1922 | |
| 1923 | if ( !dst ) |
| 1924 | { |
| 1925 | // optimization: return maximal space which could be needed for this |
| 1926 | // string instead of the exact amount which could be less if there are |
| 1927 | // any surrogates in the input |
| 1928 | // |
| 1929 | // we consider that surrogates are rare enough to make it worthwhile to |
| 1930 | // avoid running the loop below at the cost of slightly extra memory |
| 1931 | // consumption |
| 1932 | return srcLen*BYTES_PER_CHAR; |
| 1933 | } |
| 1934 | |
| 1935 | wxUint32 *outBuff = reinterpret_cast<wxUint32 *>(dst); |
| 1936 | size_t outLen = 0; |
| 1937 | for ( const wchar_t * const srcEnd = src + srcLen; src < srcEnd; ) |
| 1938 | { |
| 1939 | const wxUint32 ch = wxDecodeSurrogate(&src); |
| 1940 | if ( !src ) |
| 1941 | return wxCONV_FAILED; |
| 1942 | |
| 1943 | outLen += BYTES_PER_CHAR; |
| 1944 | |
| 1945 | if ( outLen > dstLen ) |
| 1946 | return wxCONV_FAILED; |
| 1947 | |
| 1948 | *outBuff++ = wxUINT32_SWAP_ALWAYS(ch); |
| 1949 | } |
| 1950 | |
| 1951 | return outLen; |
| 1952 | } |
| 1953 | |
| 1954 | #else // !WC_UTF16: wchar_t is UTF-32 |
| 1955 | |
| 1956 | // ---------------------------------------------------------------------------- |
| 1957 | // conversions without endianness change |
| 1958 | // ---------------------------------------------------------------------------- |
| 1959 | |
| 1960 | size_t |
| 1961 | wxMBConvUTF32straight::ToWChar(wchar_t *dst, size_t dstLen, |
| 1962 | const char *src, size_t srcLen) const |
| 1963 | { |
| 1964 | // use memcpy() as it should be much faster than hand-written loop |
| 1965 | srcLen = GetLength(src, srcLen); |
| 1966 | if ( srcLen == wxNO_LEN ) |
| 1967 | return wxCONV_FAILED; |
| 1968 | |
| 1969 | const size_t inLen = srcLen/BYTES_PER_CHAR; |
| 1970 | if ( dst ) |
| 1971 | { |
| 1972 | if ( dstLen < inLen ) |
| 1973 | return wxCONV_FAILED; |
| 1974 | |
| 1975 | memcpy(dst, src, srcLen); |
| 1976 | } |
| 1977 | |
| 1978 | return inLen; |
| 1979 | } |
| 1980 | |
| 1981 | size_t |
| 1982 | wxMBConvUTF32straight::FromWChar(char *dst, size_t dstLen, |
| 1983 | const wchar_t *src, size_t srcLen) const |
| 1984 | { |
| 1985 | if ( srcLen == wxNO_LEN ) |
| 1986 | srcLen = wxWcslen(src) + 1; |
| 1987 | |
| 1988 | srcLen *= BYTES_PER_CHAR; |
| 1989 | |
| 1990 | if ( dst ) |
| 1991 | { |
| 1992 | if ( dstLen < srcLen ) |
| 1993 | return wxCONV_FAILED; |
| 1994 | |
| 1995 | memcpy(dst, src, srcLen); |
| 1996 | } |
| 1997 | |
| 1998 | return srcLen; |
| 1999 | } |
| 2000 | |
| 2001 | // ---------------------------------------------------------------------------- |
| 2002 | // endian-reversing conversions |
| 2003 | // ---------------------------------------------------------------------------- |
| 2004 | |
| 2005 | size_t |
| 2006 | wxMBConvUTF32swap::ToWChar(wchar_t *dst, size_t dstLen, |
| 2007 | const char *src, size_t srcLen) const |
| 2008 | { |
| 2009 | srcLen = GetLength(src, srcLen); |
| 2010 | if ( srcLen == wxNO_LEN ) |
| 2011 | return wxCONV_FAILED; |
| 2012 | |
| 2013 | srcLen /= BYTES_PER_CHAR; |
| 2014 | |
| 2015 | if ( dst ) |
| 2016 | { |
| 2017 | if ( dstLen < srcLen ) |
| 2018 | return wxCONV_FAILED; |
| 2019 | |
| 2020 | const wxUint32 *inBuff = reinterpret_cast<const wxUint32 *>(src); |
| 2021 | for ( size_t n = 0; n < srcLen; n++, inBuff++ ) |
| 2022 | { |
| 2023 | *dst++ = wxUINT32_SWAP_ALWAYS(*inBuff); |
| 2024 | } |
| 2025 | } |
| 2026 | |
| 2027 | return srcLen; |
| 2028 | } |
| 2029 | |
| 2030 | size_t |
| 2031 | wxMBConvUTF32swap::FromWChar(char *dst, size_t dstLen, |
| 2032 | const wchar_t *src, size_t srcLen) const |
| 2033 | { |
| 2034 | if ( srcLen == wxNO_LEN ) |
| 2035 | srcLen = wxWcslen(src) + 1; |
| 2036 | |
| 2037 | srcLen *= BYTES_PER_CHAR; |
| 2038 | |
| 2039 | if ( dst ) |
| 2040 | { |
| 2041 | if ( dstLen < srcLen ) |
| 2042 | return wxCONV_FAILED; |
| 2043 | |
| 2044 | wxUint32 *outBuff = reinterpret_cast<wxUint32 *>(dst); |
| 2045 | for ( size_t n = 0; n < srcLen; n += BYTES_PER_CHAR, src++ ) |
| 2046 | { |
| 2047 | *outBuff++ = wxUINT32_SWAP_ALWAYS(*src); |
| 2048 | } |
| 2049 | } |
| 2050 | |
| 2051 | return srcLen; |
| 2052 | } |
| 2053 | |
| 2054 | #endif // WC_UTF16/!WC_UTF16 |
| 2055 | |
| 2056 | |
| 2057 | // ============================================================================ |
| 2058 | // The classes doing conversion using the iconv_xxx() functions |
| 2059 | // ============================================================================ |
| 2060 | |
| 2061 | #ifdef HAVE_ICONV |
| 2062 | |
| 2063 | // VS: glibc 2.1.3 is broken in that iconv() conversion to/from UCS4 fails with |
| 2064 | // E2BIG if output buffer is _exactly_ as big as needed. Such case is |
| 2065 | // (unless there's yet another bug in glibc) the only case when iconv() |
| 2066 | // returns with (size_t)-1 (which means error) and says there are 0 bytes |
| 2067 | // left in the input buffer -- when _real_ error occurs, |
| 2068 | // bytes-left-in-input buffer is non-zero. Hence, this alternative test for |
| 2069 | // iconv() failure. |
| 2070 | // [This bug does not appear in glibc 2.2.] |
| 2071 | #if defined(__GLIBC__) && __GLIBC__ == 2 && __GLIBC_MINOR__ <= 1 |
| 2072 | #define ICONV_FAILED(cres, bufLeft) ((cres == (size_t)-1) && \ |
| 2073 | (errno != E2BIG || bufLeft != 0)) |
| 2074 | #else |
| 2075 | #define ICONV_FAILED(cres, bufLeft) (cres == (size_t)-1) |
| 2076 | #endif |
| 2077 | |
| 2078 | #define ICONV_CHAR_CAST(x) ((ICONV_CONST char **)(x)) |
| 2079 | |
| 2080 | #define ICONV_T_INVALID ((iconv_t)-1) |
| 2081 | |
| 2082 | #if SIZEOF_WCHAR_T == 4 |
| 2083 | #define WC_BSWAP wxUINT32_SWAP_ALWAYS |
| 2084 | #define WC_ENC wxFONTENCODING_UTF32 |
| 2085 | #elif SIZEOF_WCHAR_T == 2 |
| 2086 | #define WC_BSWAP wxUINT16_SWAP_ALWAYS |
| 2087 | #define WC_ENC wxFONTENCODING_UTF16 |
| 2088 | #else // sizeof(wchar_t) != 2 nor 4 |
| 2089 | // does this ever happen? |
| 2090 | #error "Unknown sizeof(wchar_t): please report this to wx-dev@lists.wxwindows.org" |
| 2091 | #endif |
| 2092 | |
| 2093 | // ---------------------------------------------------------------------------- |
| 2094 | // wxMBConv_iconv: encapsulates an iconv character set |
| 2095 | // ---------------------------------------------------------------------------- |
| 2096 | |
| 2097 | class wxMBConv_iconv : public wxMBConv |
| 2098 | { |
| 2099 | public: |
| 2100 | wxMBConv_iconv(const char *name); |
| 2101 | virtual ~wxMBConv_iconv(); |
| 2102 | |
| 2103 | // implement base class virtual methods |
| 2104 | virtual size_t ToWChar(wchar_t *dst, size_t dstLen, |
| 2105 | const char *src, size_t srcLen = wxNO_LEN) const; |
| 2106 | virtual size_t FromWChar(char *dst, size_t dstLen, |
| 2107 | const wchar_t *src, size_t srcLen = wxNO_LEN) const; |
| 2108 | virtual size_t GetMBNulLen() const; |
| 2109 | |
| 2110 | #if wxUSE_UNICODE_UTF8 |
| 2111 | virtual bool IsUTF8() const; |
| 2112 | #endif |
| 2113 | |
| 2114 | virtual wxMBConv *Clone() const |
| 2115 | { |
| 2116 | wxMBConv_iconv *p = new wxMBConv_iconv(m_name); |
| 2117 | p->m_minMBCharWidth = m_minMBCharWidth; |
| 2118 | return p; |
| 2119 | } |
| 2120 | |
| 2121 | bool IsOk() const |
| 2122 | { return (m2w != ICONV_T_INVALID) && (w2m != ICONV_T_INVALID); } |
| 2123 | |
| 2124 | protected: |
| 2125 | // the iconv handlers used to translate from multibyte |
| 2126 | // to wide char and in the other direction |
| 2127 | iconv_t m2w, |
| 2128 | w2m; |
| 2129 | |
| 2130 | #if wxUSE_THREADS |
| 2131 | // guards access to m2w and w2m objects |
| 2132 | wxMutex m_iconvMutex; |
| 2133 | #endif |
| 2134 | |
| 2135 | private: |
| 2136 | // the name (for iconv_open()) of a wide char charset -- if none is |
| 2137 | // available on this machine, it will remain NULL |
| 2138 | static wxString ms_wcCharsetName; |
| 2139 | |
| 2140 | // true if the wide char encoding we use (i.e. ms_wcCharsetName) has |
| 2141 | // different endian-ness than the native one |
| 2142 | static bool ms_wcNeedsSwap; |
| 2143 | |
| 2144 | |
| 2145 | // name of the encoding handled by this conversion |
| 2146 | const char *m_name; |
| 2147 | |
| 2148 | // cached result of GetMBNulLen(); set to 0 meaning "unknown" |
| 2149 | // initially |
| 2150 | size_t m_minMBCharWidth; |
| 2151 | }; |
| 2152 | |
| 2153 | // make the constructor available for unit testing |
| 2154 | WXDLLIMPEXP_BASE wxMBConv* new_wxMBConv_iconv( const char* name ) |
| 2155 | { |
| 2156 | wxMBConv_iconv* result = new wxMBConv_iconv( name ); |
| 2157 | if ( !result->IsOk() ) |
| 2158 | { |
| 2159 | delete result; |
| 2160 | return 0; |
| 2161 | } |
| 2162 | |
| 2163 | return result; |
| 2164 | } |
| 2165 | |
| 2166 | wxString wxMBConv_iconv::ms_wcCharsetName; |
| 2167 | bool wxMBConv_iconv::ms_wcNeedsSwap = false; |
| 2168 | |
| 2169 | wxMBConv_iconv::wxMBConv_iconv(const char *name) |
| 2170 | : m_name(wxStrdup(name)) |
| 2171 | { |
| 2172 | m_minMBCharWidth = 0; |
| 2173 | |
| 2174 | // check for charset that represents wchar_t: |
| 2175 | if ( ms_wcCharsetName.empty() ) |
| 2176 | { |
| 2177 | wxLogTrace(TRACE_STRCONV, wxT("Looking for wide char codeset:")); |
| 2178 | |
| 2179 | #if wxUSE_FONTMAP |
| 2180 | const wxChar *const *names = wxFontMapperBase::GetAllEncodingNames(WC_ENC); |
| 2181 | #else // !wxUSE_FONTMAP |
| 2182 | static const wxChar *const names_static[] = |
| 2183 | { |
| 2184 | #if SIZEOF_WCHAR_T == 4 |
| 2185 | wxT("UCS-4"), |
| 2186 | #elif SIZEOF_WCHAR_T == 2 |
| 2187 | wxT("UCS-2"), |
| 2188 | #endif |
| 2189 | NULL |
| 2190 | }; |
| 2191 | const wxChar *const *names = names_static; |
| 2192 | #endif // wxUSE_FONTMAP/!wxUSE_FONTMAP |
| 2193 | |
| 2194 | for ( ; *names && ms_wcCharsetName.empty(); ++names ) |
| 2195 | { |
| 2196 | const wxString nameCS(*names); |
| 2197 | |
| 2198 | // first try charset with explicit bytesex info (e.g. "UCS-4LE"): |
| 2199 | wxString nameXE(nameCS); |
| 2200 | |
| 2201 | #ifdef WORDS_BIGENDIAN |
| 2202 | nameXE += wxT("BE"); |
| 2203 | #else // little endian |
| 2204 | nameXE += wxT("LE"); |
| 2205 | #endif |
| 2206 | |
| 2207 | wxLogTrace(TRACE_STRCONV, wxT(" trying charset \"%s\""), |
| 2208 | nameXE.c_str()); |
| 2209 | |
| 2210 | m2w = iconv_open(nameXE.ToAscii(), name); |
| 2211 | if ( m2w == ICONV_T_INVALID ) |
| 2212 | { |
| 2213 | // try charset w/o bytesex info (e.g. "UCS4") |
| 2214 | wxLogTrace(TRACE_STRCONV, wxT(" trying charset \"%s\""), |
| 2215 | nameCS.c_str()); |
| 2216 | m2w = iconv_open(nameCS.ToAscii(), name); |
| 2217 | |
| 2218 | // and check for bytesex ourselves: |
| 2219 | if ( m2w != ICONV_T_INVALID ) |
| 2220 | { |
| 2221 | char buf[2], *bufPtr; |
| 2222 | wchar_t wbuf[2]; |
| 2223 | size_t insz, outsz; |
| 2224 | size_t res; |
| 2225 | |
| 2226 | buf[0] = 'A'; |
| 2227 | buf[1] = 0; |
| 2228 | wbuf[0] = 0; |
| 2229 | insz = 2; |
| 2230 | outsz = SIZEOF_WCHAR_T * 2; |
| 2231 | char* wbufPtr = (char*)wbuf; |
| 2232 | bufPtr = buf; |
| 2233 | |
| 2234 | res = iconv( |
| 2235 | m2w, ICONV_CHAR_CAST(&bufPtr), &insz, |
| 2236 | &wbufPtr, &outsz); |
| 2237 | |
| 2238 | if (ICONV_FAILED(res, insz)) |
| 2239 | { |
| 2240 | wxLogLastError(wxT("iconv")); |
| 2241 | wxLogError(_("Conversion to charset '%s' doesn't work."), |
| 2242 | nameCS.c_str()); |
| 2243 | } |
| 2244 | else // ok, can convert to this encoding, remember it |
| 2245 | { |
| 2246 | ms_wcCharsetName = nameCS; |
| 2247 | ms_wcNeedsSwap = wbuf[0] != (wchar_t)buf[0]; |
| 2248 | } |
| 2249 | } |
| 2250 | } |
| 2251 | else // use charset not requiring byte swapping |
| 2252 | { |
| 2253 | ms_wcCharsetName = nameXE; |
| 2254 | } |
| 2255 | } |
| 2256 | |
| 2257 | wxLogTrace(TRACE_STRCONV, |
| 2258 | wxT("iconv wchar_t charset is \"%s\"%s"), |
| 2259 | ms_wcCharsetName.empty() ? wxString("<none>") |
| 2260 | : ms_wcCharsetName, |
| 2261 | ms_wcNeedsSwap ? wxT(" (needs swap)") |
| 2262 | : wxT("")); |
| 2263 | } |
| 2264 | else // we already have ms_wcCharsetName |
| 2265 | { |
| 2266 | m2w = iconv_open(ms_wcCharsetName.ToAscii(), name); |
| 2267 | } |
| 2268 | |
| 2269 | if ( ms_wcCharsetName.empty() ) |
| 2270 | { |
| 2271 | w2m = ICONV_T_INVALID; |
| 2272 | } |
| 2273 | else |
| 2274 | { |
| 2275 | w2m = iconv_open(name, ms_wcCharsetName.ToAscii()); |
| 2276 | if ( w2m == ICONV_T_INVALID ) |
| 2277 | { |
| 2278 | wxLogTrace(TRACE_STRCONV, |
| 2279 | wxT("\"%s\" -> \"%s\" works but not the converse!?"), |
| 2280 | ms_wcCharsetName.c_str(), name); |
| 2281 | } |
| 2282 | } |
| 2283 | } |
| 2284 | |
| 2285 | wxMBConv_iconv::~wxMBConv_iconv() |
| 2286 | { |
| 2287 | free(const_cast<char *>(m_name)); |
| 2288 | |
| 2289 | if ( m2w != ICONV_T_INVALID ) |
| 2290 | iconv_close(m2w); |
| 2291 | if ( w2m != ICONV_T_INVALID ) |
| 2292 | iconv_close(w2m); |
| 2293 | } |
| 2294 | |
| 2295 | size_t |
| 2296 | wxMBConv_iconv::ToWChar(wchar_t *dst, size_t dstLen, |
| 2297 | const char *src, size_t srcLen) const |
| 2298 | { |
| 2299 | if ( srcLen == wxNO_LEN ) |
| 2300 | { |
| 2301 | // find the string length: notice that must be done differently for |
| 2302 | // NUL-terminated strings and UTF-16/32 which are terminated with 2/4 |
| 2303 | // consecutive NULs |
| 2304 | const size_t nulLen = GetMBNulLen(); |
| 2305 | switch ( nulLen ) |
| 2306 | { |
| 2307 | default: |
| 2308 | return wxCONV_FAILED; |
| 2309 | |
| 2310 | case 1: |
| 2311 | srcLen = strlen(src); // arguably more optimized than our version |
| 2312 | break; |
| 2313 | |
| 2314 | case 2: |
| 2315 | case 4: |
| 2316 | // for UTF-16/32 not only we need to have 2/4 consecutive NULs |
| 2317 | // but they also have to start at character boundary and not |
| 2318 | // span two adjacent characters |
| 2319 | const char *p; |
| 2320 | for ( p = src; NotAllNULs(p, nulLen); p += nulLen ) |
| 2321 | ; |
| 2322 | srcLen = p - src; |
| 2323 | break; |
| 2324 | } |
| 2325 | |
| 2326 | // when we're determining the length of the string ourselves we count |
| 2327 | // the terminating NUL(s) as part of it and always NUL-terminate the |
| 2328 | // output |
| 2329 | srcLen += nulLen; |
| 2330 | } |
| 2331 | |
| 2332 | // we express length in the number of (wide) characters but iconv always |
| 2333 | // counts buffer sizes it in bytes |
| 2334 | dstLen *= SIZEOF_WCHAR_T; |
| 2335 | |
| 2336 | #if wxUSE_THREADS |
| 2337 | // NB: iconv() is MT-safe, but each thread must use its own iconv_t handle. |
| 2338 | // Unfortunately there are a couple of global wxCSConv objects such as |
| 2339 | // wxConvLocal that are used all over wx code, so we have to make sure |
| 2340 | // the handle is used by at most one thread at the time. Otherwise |
| 2341 | // only a few wx classes would be safe to use from non-main threads |
| 2342 | // as MB<->WC conversion would fail "randomly". |
| 2343 | wxMutexLocker lock(wxConstCast(this, wxMBConv_iconv)->m_iconvMutex); |
| 2344 | #endif // wxUSE_THREADS |
| 2345 | |
| 2346 | size_t res, cres; |
| 2347 | const char *pszPtr = src; |
| 2348 | |
| 2349 | if ( dst ) |
| 2350 | { |
| 2351 | char* bufPtr = (char*)dst; |
| 2352 | |
| 2353 | // have destination buffer, convert there |
| 2354 | size_t dstLenOrig = dstLen; |
| 2355 | cres = iconv(m2w, |
| 2356 | ICONV_CHAR_CAST(&pszPtr), &srcLen, |
| 2357 | &bufPtr, &dstLen); |
| 2358 | |
| 2359 | // convert the number of bytes converted as returned by iconv to the |
| 2360 | // number of (wide) characters converted that we need |
| 2361 | res = (dstLenOrig - dstLen) / SIZEOF_WCHAR_T; |
| 2362 | |
| 2363 | if (ms_wcNeedsSwap) |
| 2364 | { |
| 2365 | // convert to native endianness |
| 2366 | for ( unsigned i = 0; i < res; i++ ) |
| 2367 | dst[i] = WC_BSWAP(dst[i]); |
| 2368 | } |
| 2369 | } |
| 2370 | else // no destination buffer |
| 2371 | { |
| 2372 | // convert using temp buffer to calculate the size of the buffer needed |
| 2373 | wchar_t tbuf[256]; |
| 2374 | res = 0; |
| 2375 | |
| 2376 | do |
| 2377 | { |
| 2378 | char* bufPtr = (char*)tbuf; |
| 2379 | dstLen = 8 * SIZEOF_WCHAR_T; |
| 2380 | |
| 2381 | cres = iconv(m2w, |
| 2382 | ICONV_CHAR_CAST(&pszPtr), &srcLen, |
| 2383 | &bufPtr, &dstLen ); |
| 2384 | |
| 2385 | res += 8 - (dstLen / SIZEOF_WCHAR_T); |
| 2386 | } |
| 2387 | while ((cres == (size_t)-1) && (errno == E2BIG)); |
| 2388 | } |
| 2389 | |
| 2390 | if (ICONV_FAILED(cres, srcLen)) |
| 2391 | { |
| 2392 | //VS: it is ok if iconv fails, hence trace only |
| 2393 | wxLogTrace(TRACE_STRCONV, wxT("iconv failed: %s"), wxSysErrorMsg(wxSysErrorCode())); |
| 2394 | return wxCONV_FAILED; |
| 2395 | } |
| 2396 | |
| 2397 | return res; |
| 2398 | } |
| 2399 | |
| 2400 | size_t wxMBConv_iconv::FromWChar(char *dst, size_t dstLen, |
| 2401 | const wchar_t *src, size_t srcLen) const |
| 2402 | { |
| 2403 | #if wxUSE_THREADS |
| 2404 | // NB: explained in MB2WC |
| 2405 | wxMutexLocker lock(wxConstCast(this, wxMBConv_iconv)->m_iconvMutex); |
| 2406 | #endif |
| 2407 | |
| 2408 | if ( srcLen == wxNO_LEN ) |
| 2409 | srcLen = wxWcslen(src) + 1; |
| 2410 | |
| 2411 | size_t inbuflen = srcLen * SIZEOF_WCHAR_T; |
| 2412 | size_t outbuflen = dstLen; |
| 2413 | size_t res, cres; |
| 2414 | |
| 2415 | wchar_t *tmpbuf = 0; |
| 2416 | |
| 2417 | if (ms_wcNeedsSwap) |
| 2418 | { |
| 2419 | // need to copy to temp buffer to switch endianness |
| 2420 | // (doing WC_BSWAP twice on the original buffer won't work, as it |
| 2421 | // could be in read-only memory, or be accessed in some other thread) |
| 2422 | tmpbuf = (wchar_t *)malloc(inbuflen); |
| 2423 | for ( size_t i = 0; i < srcLen; i++ ) |
| 2424 | tmpbuf[i] = WC_BSWAP(src[i]); |
| 2425 | |
| 2426 | src = tmpbuf; |
| 2427 | } |
| 2428 | |
| 2429 | char* inbuf = (char*)src; |
| 2430 | if ( dst ) |
| 2431 | { |
| 2432 | // have destination buffer, convert there |
| 2433 | cres = iconv(w2m, ICONV_CHAR_CAST(&inbuf), &inbuflen, &dst, &outbuflen); |
| 2434 | |
| 2435 | res = dstLen - outbuflen; |
| 2436 | } |
| 2437 | else // no destination buffer |
| 2438 | { |
| 2439 | // convert using temp buffer to calculate the size of the buffer needed |
| 2440 | char tbuf[256]; |
| 2441 | res = 0; |
| 2442 | do |
| 2443 | { |
| 2444 | dst = tbuf; |
| 2445 | outbuflen = WXSIZEOF(tbuf); |
| 2446 | |
| 2447 | cres = iconv(w2m, ICONV_CHAR_CAST(&inbuf), &inbuflen, &dst, &outbuflen); |
| 2448 | |
| 2449 | res += WXSIZEOF(tbuf) - outbuflen; |
| 2450 | } |
| 2451 | while ((cres == (size_t)-1) && (errno == E2BIG)); |
| 2452 | } |
| 2453 | |
| 2454 | if (ms_wcNeedsSwap) |
| 2455 | { |
| 2456 | free(tmpbuf); |
| 2457 | } |
| 2458 | |
| 2459 | if (ICONV_FAILED(cres, inbuflen)) |
| 2460 | { |
| 2461 | wxLogTrace(TRACE_STRCONV, wxT("iconv failed: %s"), wxSysErrorMsg(wxSysErrorCode())); |
| 2462 | return wxCONV_FAILED; |
| 2463 | } |
| 2464 | |
| 2465 | return res; |
| 2466 | } |
| 2467 | |
| 2468 | size_t wxMBConv_iconv::GetMBNulLen() const |
| 2469 | { |
| 2470 | if ( m_minMBCharWidth == 0 ) |
| 2471 | { |
| 2472 | wxMBConv_iconv * const self = wxConstCast(this, wxMBConv_iconv); |
| 2473 | |
| 2474 | #if wxUSE_THREADS |
| 2475 | // NB: explained in MB2WC |
| 2476 | wxMutexLocker lock(self->m_iconvMutex); |
| 2477 | #endif |
| 2478 | |
| 2479 | const wchar_t *wnul = L""; |
| 2480 | char buf[8]; // should be enough for NUL in any encoding |
| 2481 | size_t inLen = sizeof(wchar_t), |
| 2482 | outLen = WXSIZEOF(buf); |
| 2483 | char *inBuff = (char *)wnul; |
| 2484 | char *outBuff = buf; |
| 2485 | if ( iconv(w2m, ICONV_CHAR_CAST(&inBuff), &inLen, &outBuff, &outLen) == (size_t)-1 ) |
| 2486 | { |
| 2487 | self->m_minMBCharWidth = (size_t)-1; |
| 2488 | } |
| 2489 | else // ok |
| 2490 | { |
| 2491 | self->m_minMBCharWidth = outBuff - buf; |
| 2492 | } |
| 2493 | } |
| 2494 | |
| 2495 | return m_minMBCharWidth; |
| 2496 | } |
| 2497 | |
| 2498 | #if wxUSE_UNICODE_UTF8 |
| 2499 | bool wxMBConv_iconv::IsUTF8() const |
| 2500 | { |
| 2501 | return wxStricmp(m_name, "UTF-8") == 0 || |
| 2502 | wxStricmp(m_name, "UTF8") == 0; |
| 2503 | } |
| 2504 | #endif |
| 2505 | |
| 2506 | #endif // HAVE_ICONV |
| 2507 | |
| 2508 | |
| 2509 | // ============================================================================ |
| 2510 | // Win32 conversion classes |
| 2511 | // ============================================================================ |
| 2512 | |
| 2513 | #ifdef wxHAVE_WIN32_MB2WC |
| 2514 | |
| 2515 | // from utils.cpp |
| 2516 | #if wxUSE_FONTMAP |
| 2517 | extern WXDLLIMPEXP_BASE long wxCharsetToCodepage(const char *charset); |
| 2518 | extern WXDLLIMPEXP_BASE long wxEncodingToCodepage(wxFontEncoding encoding); |
| 2519 | #endif |
| 2520 | |
| 2521 | class wxMBConv_win32 : public wxMBConv |
| 2522 | { |
| 2523 | public: |
| 2524 | wxMBConv_win32() |
| 2525 | { |
| 2526 | m_CodePage = CP_ACP; |
| 2527 | m_minMBCharWidth = 0; |
| 2528 | } |
| 2529 | |
| 2530 | wxMBConv_win32(const wxMBConv_win32& conv) |
| 2531 | : wxMBConv() |
| 2532 | { |
| 2533 | m_CodePage = conv.m_CodePage; |
| 2534 | m_minMBCharWidth = conv.m_minMBCharWidth; |
| 2535 | } |
| 2536 | |
| 2537 | #if wxUSE_FONTMAP |
| 2538 | wxMBConv_win32(const char* name) |
| 2539 | { |
| 2540 | m_CodePage = wxCharsetToCodepage(name); |
| 2541 | m_minMBCharWidth = 0; |
| 2542 | } |
| 2543 | |
| 2544 | wxMBConv_win32(wxFontEncoding encoding) |
| 2545 | { |
| 2546 | m_CodePage = wxEncodingToCodepage(encoding); |
| 2547 | m_minMBCharWidth = 0; |
| 2548 | } |
| 2549 | #endif // wxUSE_FONTMAP |
| 2550 | |
| 2551 | virtual size_t MB2WC(wchar_t *buf, const char *psz, size_t n) const |
| 2552 | { |
| 2553 | // note that we have to use MB_ERR_INVALID_CHARS flag as it without it |
| 2554 | // the behaviour is not compatible with the Unix version (using iconv) |
| 2555 | // and break the library itself, e.g. wxTextInputStream::NextChar() |
| 2556 | // wouldn't work if reading an incomplete MB char didn't result in an |
| 2557 | // error |
| 2558 | // |
| 2559 | // Moreover, MB_ERR_INVALID_CHARS is only supported on Win 2K SP4 or |
| 2560 | // Win XP or newer and it is not supported for UTF-[78] so we always |
| 2561 | // use our own conversions in this case. See |
| 2562 | // http://blogs.msdn.com/michkap/archive/2005/04/19/409566.aspx |
| 2563 | // http://msdn.microsoft.com/library/en-us/intl/unicode_17si.asp |
| 2564 | if ( m_CodePage == CP_UTF8 ) |
| 2565 | { |
| 2566 | return wxMBConvUTF8().MB2WC(buf, psz, n); |
| 2567 | } |
| 2568 | |
| 2569 | if ( m_CodePage == CP_UTF7 ) |
| 2570 | { |
| 2571 | return wxMBConvUTF7().MB2WC(buf, psz, n); |
| 2572 | } |
| 2573 | |
| 2574 | int flags = 0; |
| 2575 | if ( (m_CodePage < 50000 && m_CodePage != CP_SYMBOL) && |
| 2576 | IsAtLeastWin2kSP4() ) |
| 2577 | { |
| 2578 | flags = MB_ERR_INVALID_CHARS; |
| 2579 | } |
| 2580 | |
| 2581 | const size_t len = ::MultiByteToWideChar |
| 2582 | ( |
| 2583 | m_CodePage, // code page |
| 2584 | flags, // flags: fall on error |
| 2585 | psz, // input string |
| 2586 | -1, // its length (NUL-terminated) |
| 2587 | buf, // output string |
| 2588 | buf ? n : 0 // size of output buffer |
| 2589 | ); |
| 2590 | if ( !len ) |
| 2591 | { |
| 2592 | // function totally failed |
| 2593 | return wxCONV_FAILED; |
| 2594 | } |
| 2595 | |
| 2596 | // if we were really converting and didn't use MB_ERR_INVALID_CHARS, |
| 2597 | // check if we succeeded, by doing a double trip: |
| 2598 | if ( !flags && buf ) |
| 2599 | { |
| 2600 | const size_t mbLen = strlen(psz); |
| 2601 | wxCharBuffer mbBuf(mbLen); |
| 2602 | if ( ::WideCharToMultiByte |
| 2603 | ( |
| 2604 | m_CodePage, |
| 2605 | 0, |
| 2606 | buf, |
| 2607 | -1, |
| 2608 | mbBuf.data(), |
| 2609 | mbLen + 1, // size in bytes, not length |
| 2610 | NULL, |
| 2611 | NULL |
| 2612 | ) == 0 || |
| 2613 | strcmp(mbBuf, psz) != 0 ) |
| 2614 | { |
| 2615 | // we didn't obtain the same thing we started from, hence |
| 2616 | // the conversion was lossy and we consider that it failed |
| 2617 | return wxCONV_FAILED; |
| 2618 | } |
| 2619 | } |
| 2620 | |
| 2621 | // note that it returns count of written chars for buf != NULL and size |
| 2622 | // of the needed buffer for buf == NULL so in either case the length of |
| 2623 | // the string (which never includes the terminating NUL) is one less |
| 2624 | return len - 1; |
| 2625 | } |
| 2626 | |
| 2627 | virtual size_t WC2MB(char *buf, const wchar_t *pwz, size_t n) const |
| 2628 | { |
| 2629 | /* |
| 2630 | we have a problem here: by default, WideCharToMultiByte() may |
| 2631 | replace characters unrepresentable in the target code page with bad |
| 2632 | quality approximations such as turning "1/2" symbol (U+00BD) into |
| 2633 | "1" for the code pages which don't have it and we, obviously, want |
| 2634 | to avoid this at any price |
| 2635 | |
| 2636 | the trouble is that this function does it _silently_, i.e. it won't |
| 2637 | even tell us whether it did or not... Win98/2000 and higher provide |
| 2638 | WC_NO_BEST_FIT_CHARS but it doesn't work for the older systems and |
| 2639 | we have to resort to a round trip, i.e. check that converting back |
| 2640 | results in the same string -- this is, of course, expensive but |
| 2641 | otherwise we simply can't be sure to not garble the data. |
| 2642 | */ |
| 2643 | |
| 2644 | // determine if we can rely on WC_NO_BEST_FIT_CHARS: according to MSDN |
| 2645 | // it doesn't work with CJK encodings (which we test for rather roughly |
| 2646 | // here...) nor with UTF-7/8 nor, of course, with Windows versions not |
| 2647 | // supporting it |
| 2648 | BOOL usedDef wxDUMMY_INITIALIZE(false); |
| 2649 | BOOL *pUsedDef; |
| 2650 | int flags; |
| 2651 | if ( CanUseNoBestFit() && m_CodePage < 50000 ) |
| 2652 | { |
| 2653 | // it's our lucky day |
| 2654 | flags = WC_NO_BEST_FIT_CHARS; |
| 2655 | pUsedDef = &usedDef; |
| 2656 | } |
| 2657 | else // old system or unsupported encoding |
| 2658 | { |
| 2659 | flags = 0; |
| 2660 | pUsedDef = NULL; |
| 2661 | } |
| 2662 | |
| 2663 | const size_t len = ::WideCharToMultiByte |
| 2664 | ( |
| 2665 | m_CodePage, // code page |
| 2666 | flags, // either none or no best fit |
| 2667 | pwz, // input string |
| 2668 | -1, // it is (wide) NUL-terminated |
| 2669 | buf, // output buffer |
| 2670 | buf ? n : 0, // and its size |
| 2671 | NULL, // default "replacement" char |
| 2672 | pUsedDef // [out] was it used? |
| 2673 | ); |
| 2674 | |
| 2675 | if ( !len ) |
| 2676 | { |
| 2677 | // function totally failed |
| 2678 | return wxCONV_FAILED; |
| 2679 | } |
| 2680 | |
| 2681 | // we did something, check if we really succeeded |
| 2682 | if ( flags ) |
| 2683 | { |
| 2684 | // check if the conversion failed, i.e. if any replacements |
| 2685 | // were done |
| 2686 | if ( usedDef ) |
| 2687 | return wxCONV_FAILED; |
| 2688 | } |
| 2689 | else // we must resort to double tripping... |
| 2690 | { |
| 2691 | // first we need to ensure that we really have the MB data: this is |
| 2692 | // not the case if we're called with NULL buffer, in which case we |
| 2693 | // need to do the conversion yet again |
| 2694 | wxCharBuffer bufDef; |
| 2695 | if ( !buf ) |
| 2696 | { |
| 2697 | bufDef = wxCharBuffer(len); |
| 2698 | buf = bufDef.data(); |
| 2699 | if ( !::WideCharToMultiByte(m_CodePage, flags, pwz, -1, |
| 2700 | buf, len, NULL, NULL) ) |
| 2701 | return wxCONV_FAILED; |
| 2702 | } |
| 2703 | |
| 2704 | if ( !n ) |
| 2705 | n = wcslen(pwz); |
| 2706 | wxWCharBuffer wcBuf(n); |
| 2707 | if ( MB2WC(wcBuf.data(), buf, n + 1) == wxCONV_FAILED || |
| 2708 | wcscmp(wcBuf, pwz) != 0 ) |
| 2709 | { |
| 2710 | // we didn't obtain the same thing we started from, hence |
| 2711 | // the conversion was lossy and we consider that it failed |
| 2712 | return wxCONV_FAILED; |
| 2713 | } |
| 2714 | } |
| 2715 | |
| 2716 | // see the comment above for the reason of "len - 1" |
| 2717 | return len - 1; |
| 2718 | } |
| 2719 | |
| 2720 | virtual size_t GetMBNulLen() const |
| 2721 | { |
| 2722 | if ( m_minMBCharWidth == 0 ) |
| 2723 | { |
| 2724 | int len = ::WideCharToMultiByte |
| 2725 | ( |
| 2726 | m_CodePage, // code page |
| 2727 | 0, // no flags |
| 2728 | L"", // input string |
| 2729 | 1, // translate just the NUL |
| 2730 | NULL, // output buffer |
| 2731 | 0, // and its size |
| 2732 | NULL, // no replacement char |
| 2733 | NULL // [out] don't care if it was used |
| 2734 | ); |
| 2735 | |
| 2736 | wxMBConv_win32 * const self = wxConstCast(this, wxMBConv_win32); |
| 2737 | switch ( len ) |
| 2738 | { |
| 2739 | default: |
| 2740 | wxLogDebug(wxT("Unexpected NUL length %d"), len); |
| 2741 | self->m_minMBCharWidth = (size_t)-1; |
| 2742 | break; |
| 2743 | |
| 2744 | case 0: |
| 2745 | self->m_minMBCharWidth = (size_t)-1; |
| 2746 | break; |
| 2747 | |
| 2748 | case 1: |
| 2749 | case 2: |
| 2750 | case 4: |
| 2751 | self->m_minMBCharWidth = len; |
| 2752 | break; |
| 2753 | } |
| 2754 | } |
| 2755 | |
| 2756 | return m_minMBCharWidth; |
| 2757 | } |
| 2758 | |
| 2759 | virtual wxMBConv *Clone() const { return new wxMBConv_win32(*this); } |
| 2760 | |
| 2761 | bool IsOk() const { return m_CodePage != -1; } |
| 2762 | |
| 2763 | private: |
| 2764 | static bool CanUseNoBestFit() |
| 2765 | { |
| 2766 | static int s_isWin98Or2k = -1; |
| 2767 | |
| 2768 | if ( s_isWin98Or2k == -1 ) |
| 2769 | { |
| 2770 | int verMaj, verMin; |
| 2771 | switch ( wxGetOsVersion(&verMaj, &verMin) ) |
| 2772 | { |
| 2773 | case wxOS_WINDOWS_9X: |
| 2774 | s_isWin98Or2k = verMaj >= 4 && verMin >= 10; |
| 2775 | break; |
| 2776 | |
| 2777 | case wxOS_WINDOWS_NT: |
| 2778 | s_isWin98Or2k = verMaj >= 5; |
| 2779 | break; |
| 2780 | |
| 2781 | default: |
| 2782 | // unknown: be conservative by default |
| 2783 | s_isWin98Or2k = 0; |
| 2784 | break; |
| 2785 | } |
| 2786 | |
| 2787 | wxASSERT_MSG( s_isWin98Or2k != -1, wxT("should be set above") ); |
| 2788 | } |
| 2789 | |
| 2790 | return s_isWin98Or2k == 1; |
| 2791 | } |
| 2792 | |
| 2793 | static bool IsAtLeastWin2kSP4() |
| 2794 | { |
| 2795 | #ifdef __WXWINCE__ |
| 2796 | return false; |
| 2797 | #else |
| 2798 | static int s_isAtLeastWin2kSP4 = -1; |
| 2799 | |
| 2800 | if ( s_isAtLeastWin2kSP4 == -1 ) |
| 2801 | { |
| 2802 | OSVERSIONINFOEX ver; |
| 2803 | |
| 2804 | memset(&ver, 0, sizeof(ver)); |
| 2805 | ver.dwOSVersionInfoSize = sizeof(ver); |
| 2806 | GetVersionEx((OSVERSIONINFO*)&ver); |
| 2807 | |
| 2808 | s_isAtLeastWin2kSP4 = |
| 2809 | ((ver.dwMajorVersion > 5) || // Vista+ |
| 2810 | (ver.dwMajorVersion == 5 && ver.dwMinorVersion > 0) || // XP/2003 |
| 2811 | (ver.dwMajorVersion == 5 && ver.dwMinorVersion == 0 && |
| 2812 | ver.wServicePackMajor >= 4)) // 2000 SP4+ |
| 2813 | ? 1 : 0; |
| 2814 | } |
| 2815 | |
| 2816 | return s_isAtLeastWin2kSP4 == 1; |
| 2817 | #endif |
| 2818 | } |
| 2819 | |
| 2820 | |
| 2821 | // the code page we're working with |
| 2822 | long m_CodePage; |
| 2823 | |
| 2824 | // cached result of GetMBNulLen(), set to 0 initially meaning |
| 2825 | // "unknown" |
| 2826 | size_t m_minMBCharWidth; |
| 2827 | }; |
| 2828 | |
| 2829 | #endif // wxHAVE_WIN32_MB2WC |
| 2830 | |
| 2831 | |
| 2832 | // ============================================================================ |
| 2833 | // wxEncodingConverter based conversion classes |
| 2834 | // ============================================================================ |
| 2835 | |
| 2836 | #if wxUSE_FONTMAP |
| 2837 | |
| 2838 | class wxMBConv_wxwin : public wxMBConv |
| 2839 | { |
| 2840 | private: |
| 2841 | void Init() |
| 2842 | { |
| 2843 | // Refuse to use broken wxEncodingConverter code for Mac-specific encodings. |
| 2844 | // The wxMBConv_cf class does a better job. |
| 2845 | m_ok = (m_enc < wxFONTENCODING_MACMIN || m_enc > wxFONTENCODING_MACMAX) && |
| 2846 | m2w.Init(m_enc, wxFONTENCODING_UNICODE) && |
| 2847 | w2m.Init(wxFONTENCODING_UNICODE, m_enc); |
| 2848 | } |
| 2849 | |
| 2850 | public: |
| 2851 | // temporarily just use wxEncodingConverter stuff, |
| 2852 | // so that it works while a better implementation is built |
| 2853 | wxMBConv_wxwin(const char* name) |
| 2854 | { |
| 2855 | if (name) |
| 2856 | m_enc = wxFontMapperBase::Get()->CharsetToEncoding(name, false); |
| 2857 | else |
| 2858 | m_enc = wxFONTENCODING_SYSTEM; |
| 2859 | |
| 2860 | Init(); |
| 2861 | } |
| 2862 | |
| 2863 | wxMBConv_wxwin(wxFontEncoding enc) |
| 2864 | { |
| 2865 | m_enc = enc; |
| 2866 | |
| 2867 | Init(); |
| 2868 | } |
| 2869 | |
| 2870 | size_t MB2WC(wchar_t *buf, const char *psz, size_t WXUNUSED(n)) const |
| 2871 | { |
| 2872 | size_t inbuf = strlen(psz); |
| 2873 | if (buf) |
| 2874 | { |
| 2875 | if (!m2w.Convert(psz, buf)) |
| 2876 | return wxCONV_FAILED; |
| 2877 | } |
| 2878 | return inbuf; |
| 2879 | } |
| 2880 | |
| 2881 | size_t WC2MB(char *buf, const wchar_t *psz, size_t WXUNUSED(n)) const |
| 2882 | { |
| 2883 | const size_t inbuf = wxWcslen(psz); |
| 2884 | if (buf) |
| 2885 | { |
| 2886 | if (!w2m.Convert(psz, buf)) |
| 2887 | return wxCONV_FAILED; |
| 2888 | } |
| 2889 | |
| 2890 | return inbuf; |
| 2891 | } |
| 2892 | |
| 2893 | virtual size_t GetMBNulLen() const |
| 2894 | { |
| 2895 | switch ( m_enc ) |
| 2896 | { |
| 2897 | case wxFONTENCODING_UTF16BE: |
| 2898 | case wxFONTENCODING_UTF16LE: |
| 2899 | return 2; |
| 2900 | |
| 2901 | case wxFONTENCODING_UTF32BE: |
| 2902 | case wxFONTENCODING_UTF32LE: |
| 2903 | return 4; |
| 2904 | |
| 2905 | default: |
| 2906 | return 1; |
| 2907 | } |
| 2908 | } |
| 2909 | |
| 2910 | virtual wxMBConv *Clone() const { return new wxMBConv_wxwin(m_enc); } |
| 2911 | |
| 2912 | bool IsOk() const { return m_ok; } |
| 2913 | |
| 2914 | public: |
| 2915 | wxFontEncoding m_enc; |
| 2916 | wxEncodingConverter m2w, w2m; |
| 2917 | |
| 2918 | private: |
| 2919 | // were we initialized successfully? |
| 2920 | bool m_ok; |
| 2921 | |
| 2922 | wxDECLARE_NO_COPY_CLASS(wxMBConv_wxwin); |
| 2923 | }; |
| 2924 | |
| 2925 | // make the constructors available for unit testing |
| 2926 | WXDLLIMPEXP_BASE wxMBConv* new_wxMBConv_wxwin( const char* name ) |
| 2927 | { |
| 2928 | wxMBConv_wxwin* result = new wxMBConv_wxwin( name ); |
| 2929 | if ( !result->IsOk() ) |
| 2930 | { |
| 2931 | delete result; |
| 2932 | return 0; |
| 2933 | } |
| 2934 | |
| 2935 | return result; |
| 2936 | } |
| 2937 | |
| 2938 | #endif // wxUSE_FONTMAP |
| 2939 | |
| 2940 | // ============================================================================ |
| 2941 | // wxCSConv implementation |
| 2942 | // ============================================================================ |
| 2943 | |
| 2944 | void wxCSConv::Init() |
| 2945 | { |
| 2946 | m_name = NULL; |
| 2947 | m_convReal = NULL; |
| 2948 | } |
| 2949 | |
| 2950 | void wxCSConv::SetEncoding(wxFontEncoding encoding) |
| 2951 | { |
| 2952 | switch ( encoding ) |
| 2953 | { |
| 2954 | case wxFONTENCODING_MAX: |
| 2955 | case wxFONTENCODING_SYSTEM: |
| 2956 | if ( m_name ) |
| 2957 | { |
| 2958 | // It's ok to not have encoding value if we have a name for it. |
| 2959 | m_encoding = wxFONTENCODING_SYSTEM; |
| 2960 | } |
| 2961 | else // No name neither. |
| 2962 | { |
| 2963 | // Fall back to the system default encoding in this case (not |
| 2964 | // sure how much sense does this make but this is how the old |
| 2965 | // code used to behave). |
| 2966 | #if wxUSE_INTL |
| 2967 | m_encoding = wxLocale::GetSystemEncoding(); |
| 2968 | if ( m_encoding == wxFONTENCODING_SYSTEM ) |
| 2969 | #endif // wxUSE_INTL |
| 2970 | m_encoding = wxFONTENCODING_ISO8859_1; |
| 2971 | } |
| 2972 | break; |
| 2973 | |
| 2974 | case wxFONTENCODING_DEFAULT: |
| 2975 | // wxFONTENCODING_DEFAULT is same as US-ASCII in this context |
| 2976 | m_encoding = wxFONTENCODING_ISO8859_1; |
| 2977 | break; |
| 2978 | |
| 2979 | default: |
| 2980 | // Just use the provided encoding. |
| 2981 | m_encoding = encoding; |
| 2982 | } |
| 2983 | } |
| 2984 | |
| 2985 | wxCSConv::wxCSConv(const wxString& charset) |
| 2986 | { |
| 2987 | Init(); |
| 2988 | |
| 2989 | if ( !charset.empty() ) |
| 2990 | { |
| 2991 | SetName(charset.ToAscii()); |
| 2992 | } |
| 2993 | |
| 2994 | #if wxUSE_FONTMAP |
| 2995 | SetEncoding(wxFontMapperBase::GetEncodingFromName(charset)); |
| 2996 | #else |
| 2997 | SetEncoding(wxFONTENCODING_SYSTEM); |
| 2998 | #endif |
| 2999 | |
| 3000 | m_convReal = DoCreate(); |
| 3001 | } |
| 3002 | |
| 3003 | wxCSConv::wxCSConv(wxFontEncoding encoding) |
| 3004 | { |
| 3005 | if ( encoding == wxFONTENCODING_MAX || encoding == wxFONTENCODING_DEFAULT ) |
| 3006 | { |
| 3007 | wxFAIL_MSG( wxT("invalid encoding value in wxCSConv ctor") ); |
| 3008 | |
| 3009 | encoding = wxFONTENCODING_SYSTEM; |
| 3010 | } |
| 3011 | |
| 3012 | Init(); |
| 3013 | |
| 3014 | SetEncoding(encoding); |
| 3015 | |
| 3016 | m_convReal = DoCreate(); |
| 3017 | } |
| 3018 | |
| 3019 | wxCSConv::~wxCSConv() |
| 3020 | { |
| 3021 | Clear(); |
| 3022 | } |
| 3023 | |
| 3024 | wxCSConv::wxCSConv(const wxCSConv& conv) |
| 3025 | : wxMBConv() |
| 3026 | { |
| 3027 | Init(); |
| 3028 | |
| 3029 | SetName(conv.m_name); |
| 3030 | SetEncoding(conv.m_encoding); |
| 3031 | |
| 3032 | m_convReal = DoCreate(); |
| 3033 | } |
| 3034 | |
| 3035 | wxCSConv& wxCSConv::operator=(const wxCSConv& conv) |
| 3036 | { |
| 3037 | Clear(); |
| 3038 | |
| 3039 | SetName(conv.m_name); |
| 3040 | SetEncoding(conv.m_encoding); |
| 3041 | |
| 3042 | m_convReal = DoCreate(); |
| 3043 | |
| 3044 | return *this; |
| 3045 | } |
| 3046 | |
| 3047 | void wxCSConv::Clear() |
| 3048 | { |
| 3049 | free(m_name); |
| 3050 | m_name = NULL; |
| 3051 | |
| 3052 | wxDELETE(m_convReal); |
| 3053 | } |
| 3054 | |
| 3055 | void wxCSConv::SetName(const char *charset) |
| 3056 | { |
| 3057 | if ( charset ) |
| 3058 | m_name = wxStrdup(charset); |
| 3059 | } |
| 3060 | |
| 3061 | #if wxUSE_FONTMAP |
| 3062 | |
| 3063 | WX_DECLARE_HASH_MAP( wxFontEncoding, wxString, wxIntegerHash, wxIntegerEqual, |
| 3064 | wxEncodingNameCache ); |
| 3065 | |
| 3066 | static wxEncodingNameCache gs_nameCache; |
| 3067 | #endif |
| 3068 | |
| 3069 | wxMBConv *wxCSConv::DoCreate() const |
| 3070 | { |
| 3071 | #if wxUSE_FONTMAP |
| 3072 | wxLogTrace(TRACE_STRCONV, |
| 3073 | wxT("creating conversion for %s"), |
| 3074 | (m_name ? m_name |
| 3075 | : (const char*)wxFontMapperBase::GetEncodingName(m_encoding).mb_str())); |
| 3076 | #endif // wxUSE_FONTMAP |
| 3077 | |
| 3078 | // check for the special case of ASCII or ISO8859-1 charset: as we have |
| 3079 | // special knowledge of it anyhow, we don't need to create a special |
| 3080 | // conversion object |
| 3081 | if ( m_encoding == wxFONTENCODING_ISO8859_1 ) |
| 3082 | { |
| 3083 | // don't convert at all |
| 3084 | return NULL; |
| 3085 | } |
| 3086 | |
| 3087 | // we trust OS to do conversion better than we can so try external |
| 3088 | // conversion methods first |
| 3089 | // |
| 3090 | // the full order is: |
| 3091 | // 1. OS conversion (iconv() under Unix or Win32 API) |
| 3092 | // 2. hard coded conversions for UTF |
| 3093 | // 3. wxEncodingConverter as fall back |
| 3094 | |
| 3095 | // step (1) |
| 3096 | #ifdef HAVE_ICONV |
| 3097 | #if !wxUSE_FONTMAP |
| 3098 | if ( m_name ) |
| 3099 | #endif // !wxUSE_FONTMAP |
| 3100 | { |
| 3101 | #if wxUSE_FONTMAP |
| 3102 | wxFontEncoding encoding(m_encoding); |
| 3103 | #endif |
| 3104 | |
| 3105 | if ( m_name ) |
| 3106 | { |
| 3107 | wxMBConv_iconv *conv = new wxMBConv_iconv(m_name); |
| 3108 | if ( conv->IsOk() ) |
| 3109 | return conv; |
| 3110 | |
| 3111 | delete conv; |
| 3112 | |
| 3113 | #if wxUSE_FONTMAP |
| 3114 | encoding = |
| 3115 | wxFontMapperBase::Get()->CharsetToEncoding(m_name, false); |
| 3116 | #endif // wxUSE_FONTMAP |
| 3117 | } |
| 3118 | #if wxUSE_FONTMAP |
| 3119 | { |
| 3120 | const wxEncodingNameCache::iterator it = gs_nameCache.find(encoding); |
| 3121 | if ( it != gs_nameCache.end() ) |
| 3122 | { |
| 3123 | if ( it->second.empty() ) |
| 3124 | return NULL; |
| 3125 | |
| 3126 | wxMBConv_iconv *conv = new wxMBConv_iconv(it->second.ToAscii()); |
| 3127 | if ( conv->IsOk() ) |
| 3128 | return conv; |
| 3129 | |
| 3130 | delete conv; |
| 3131 | } |
| 3132 | |
| 3133 | const wxChar* const* names = wxFontMapperBase::GetAllEncodingNames(encoding); |
| 3134 | // CS : in case this does not return valid names (eg for MacRoman) |
| 3135 | // encoding got a 'failure' entry in the cache all the same, |
| 3136 | // although it just has to be created using a different method, so |
| 3137 | // only store failed iconv creation attempts (or perhaps we |
| 3138 | // shoulnd't do this at all ?) |
| 3139 | if ( names[0] != NULL ) |
| 3140 | { |
| 3141 | for ( ; *names; ++names ) |
| 3142 | { |
| 3143 | // FIXME-UTF8: wxFontMapperBase::GetAllEncodingNames() |
| 3144 | // will need changes that will obsolete this |
| 3145 | wxString name(*names); |
| 3146 | wxMBConv_iconv *conv = new wxMBConv_iconv(name.ToAscii()); |
| 3147 | if ( conv->IsOk() ) |
| 3148 | { |
| 3149 | gs_nameCache[encoding] = *names; |
| 3150 | return conv; |
| 3151 | } |
| 3152 | |
| 3153 | delete conv; |
| 3154 | } |
| 3155 | |
| 3156 | gs_nameCache[encoding] = wxT(""); // cache the failure |
| 3157 | } |
| 3158 | } |
| 3159 | #endif // wxUSE_FONTMAP |
| 3160 | } |
| 3161 | #endif // HAVE_ICONV |
| 3162 | |
| 3163 | #ifdef wxHAVE_WIN32_MB2WC |
| 3164 | { |
| 3165 | #if wxUSE_FONTMAP |
| 3166 | wxMBConv_win32 *conv = m_name ? new wxMBConv_win32(m_name) |
| 3167 | : new wxMBConv_win32(m_encoding); |
| 3168 | if ( conv->IsOk() ) |
| 3169 | return conv; |
| 3170 | |
| 3171 | delete conv; |
| 3172 | #else |
| 3173 | return NULL; |
| 3174 | #endif |
| 3175 | } |
| 3176 | #endif // wxHAVE_WIN32_MB2WC |
| 3177 | |
| 3178 | #ifdef __DARWIN__ |
| 3179 | { |
| 3180 | // leave UTF16 and UTF32 to the built-ins of wx |
| 3181 | if ( m_name || ( m_encoding < wxFONTENCODING_UTF16BE || |
| 3182 | ( m_encoding >= wxFONTENCODING_MACMIN && m_encoding <= wxFONTENCODING_MACMAX ) ) ) |
| 3183 | { |
| 3184 | #if wxUSE_FONTMAP |
| 3185 | wxMBConv_cf *conv = m_name ? new wxMBConv_cf(m_name) |
| 3186 | : new wxMBConv_cf(m_encoding); |
| 3187 | #else |
| 3188 | wxMBConv_cf *conv = new wxMBConv_cf(m_encoding); |
| 3189 | #endif |
| 3190 | |
| 3191 | if ( conv->IsOk() ) |
| 3192 | return conv; |
| 3193 | |
| 3194 | delete conv; |
| 3195 | } |
| 3196 | } |
| 3197 | #endif // __DARWIN__ |
| 3198 | |
| 3199 | // step (2) |
| 3200 | wxFontEncoding enc = m_encoding; |
| 3201 | #if wxUSE_FONTMAP |
| 3202 | if ( enc == wxFONTENCODING_SYSTEM && m_name ) |
| 3203 | { |
| 3204 | // use "false" to suppress interactive dialogs -- we can be called from |
| 3205 | // anywhere and popping up a dialog from here is the last thing we want to |
| 3206 | // do |
| 3207 | enc = wxFontMapperBase::Get()->CharsetToEncoding(m_name, false); |
| 3208 | } |
| 3209 | #endif // wxUSE_FONTMAP |
| 3210 | |
| 3211 | switch ( enc ) |
| 3212 | { |
| 3213 | case wxFONTENCODING_UTF7: |
| 3214 | return new wxMBConvUTF7; |
| 3215 | |
| 3216 | case wxFONTENCODING_UTF8: |
| 3217 | return new wxMBConvUTF8; |
| 3218 | |
| 3219 | case wxFONTENCODING_UTF16BE: |
| 3220 | return new wxMBConvUTF16BE; |
| 3221 | |
| 3222 | case wxFONTENCODING_UTF16LE: |
| 3223 | return new wxMBConvUTF16LE; |
| 3224 | |
| 3225 | case wxFONTENCODING_UTF32BE: |
| 3226 | return new wxMBConvUTF32BE; |
| 3227 | |
| 3228 | case wxFONTENCODING_UTF32LE: |
| 3229 | return new wxMBConvUTF32LE; |
| 3230 | |
| 3231 | default: |
| 3232 | // nothing to do but put here to suppress gcc warnings |
| 3233 | break; |
| 3234 | } |
| 3235 | |
| 3236 | // step (3) |
| 3237 | #if wxUSE_FONTMAP |
| 3238 | { |
| 3239 | wxMBConv_wxwin *conv = m_name ? new wxMBConv_wxwin(m_name) |
| 3240 | : new wxMBConv_wxwin(m_encoding); |
| 3241 | if ( conv->IsOk() ) |
| 3242 | return conv; |
| 3243 | |
| 3244 | delete conv; |
| 3245 | } |
| 3246 | |
| 3247 | wxLogTrace(TRACE_STRCONV, |
| 3248 | wxT("encoding \"%s\" is not supported by this system"), |
| 3249 | (m_name ? wxString(m_name) |
| 3250 | : wxFontMapperBase::GetEncodingName(m_encoding))); |
| 3251 | #endif // wxUSE_FONTMAP |
| 3252 | |
| 3253 | return NULL; |
| 3254 | } |
| 3255 | |
| 3256 | bool wxCSConv::IsOk() const |
| 3257 | { |
| 3258 | // special case: no convReal created for wxFONTENCODING_ISO8859_1 |
| 3259 | if ( m_encoding == wxFONTENCODING_ISO8859_1 ) |
| 3260 | return true; // always ok as we do it ourselves |
| 3261 | |
| 3262 | // m_convReal->IsOk() is called at its own creation, so we know it must |
| 3263 | // be ok if m_convReal is non-NULL |
| 3264 | return m_convReal != NULL; |
| 3265 | } |
| 3266 | |
| 3267 | size_t wxCSConv::ToWChar(wchar_t *dst, size_t dstLen, |
| 3268 | const char *src, size_t srcLen) const |
| 3269 | { |
| 3270 | if (m_convReal) |
| 3271 | return m_convReal->ToWChar(dst, dstLen, src, srcLen); |
| 3272 | |
| 3273 | // latin-1 (direct) |
| 3274 | if ( srcLen == wxNO_LEN ) |
| 3275 | srcLen = strlen(src) + 1; // take trailing NUL too |
| 3276 | |
| 3277 | if ( dst ) |
| 3278 | { |
| 3279 | if ( dstLen < srcLen ) |
| 3280 | return wxCONV_FAILED; |
| 3281 | |
| 3282 | for ( size_t n = 0; n < srcLen; n++ ) |
| 3283 | dst[n] = (unsigned char)(src[n]); |
| 3284 | } |
| 3285 | |
| 3286 | return srcLen; |
| 3287 | } |
| 3288 | |
| 3289 | size_t wxCSConv::FromWChar(char *dst, size_t dstLen, |
| 3290 | const wchar_t *src, size_t srcLen) const |
| 3291 | { |
| 3292 | if (m_convReal) |
| 3293 | return m_convReal->FromWChar(dst, dstLen, src, srcLen); |
| 3294 | |
| 3295 | // latin-1 (direct) |
| 3296 | if ( srcLen == wxNO_LEN ) |
| 3297 | srcLen = wxWcslen(src) + 1; |
| 3298 | |
| 3299 | if ( dst ) |
| 3300 | { |
| 3301 | if ( dstLen < srcLen ) |
| 3302 | return wxCONV_FAILED; |
| 3303 | |
| 3304 | for ( size_t n = 0; n < srcLen; n++ ) |
| 3305 | { |
| 3306 | if ( src[n] > 0xFF ) |
| 3307 | return wxCONV_FAILED; |
| 3308 | |
| 3309 | dst[n] = (char)src[n]; |
| 3310 | } |
| 3311 | |
| 3312 | } |
| 3313 | else // still need to check the input validity |
| 3314 | { |
| 3315 | for ( size_t n = 0; n < srcLen; n++ ) |
| 3316 | { |
| 3317 | if ( src[n] > 0xFF ) |
| 3318 | return wxCONV_FAILED; |
| 3319 | } |
| 3320 | } |
| 3321 | |
| 3322 | return srcLen; |
| 3323 | } |
| 3324 | |
| 3325 | size_t wxCSConv::GetMBNulLen() const |
| 3326 | { |
| 3327 | if ( m_convReal ) |
| 3328 | return m_convReal->GetMBNulLen(); |
| 3329 | |
| 3330 | // otherwise, we are ISO-8859-1 |
| 3331 | return 1; |
| 3332 | } |
| 3333 | |
| 3334 | #if wxUSE_UNICODE_UTF8 |
| 3335 | bool wxCSConv::IsUTF8() const |
| 3336 | { |
| 3337 | if ( m_convReal ) |
| 3338 | return m_convReal->IsUTF8(); |
| 3339 | |
| 3340 | // otherwise, we are ISO-8859-1 |
| 3341 | return false; |
| 3342 | } |
| 3343 | #endif |
| 3344 | |
| 3345 | |
| 3346 | #if wxUSE_UNICODE |
| 3347 | |
| 3348 | wxWCharBuffer wxSafeConvertMB2WX(const char *s) |
| 3349 | { |
| 3350 | if ( !s ) |
| 3351 | return wxWCharBuffer(); |
| 3352 | |
| 3353 | wxWCharBuffer wbuf(wxConvLibc.cMB2WX(s)); |
| 3354 | if ( !wbuf ) |
| 3355 | wbuf = wxMBConvUTF8().cMB2WX(s); |
| 3356 | if ( !wbuf ) |
| 3357 | wbuf = wxConvISO8859_1.cMB2WX(s); |
| 3358 | |
| 3359 | return wbuf; |
| 3360 | } |
| 3361 | |
| 3362 | wxCharBuffer wxSafeConvertWX2MB(const wchar_t *ws) |
| 3363 | { |
| 3364 | if ( !ws ) |
| 3365 | return wxCharBuffer(); |
| 3366 | |
| 3367 | wxCharBuffer buf(wxConvLibc.cWX2MB(ws)); |
| 3368 | if ( !buf ) |
| 3369 | buf = wxMBConvUTF8(wxMBConvUTF8::MAP_INVALID_UTF8_TO_OCTAL).cWX2MB(ws); |
| 3370 | |
| 3371 | return buf; |
| 3372 | } |
| 3373 | |
| 3374 | #endif // wxUSE_UNICODE |
| 3375 | |
| 3376 | // ---------------------------------------------------------------------------- |
| 3377 | // globals |
| 3378 | // ---------------------------------------------------------------------------- |
| 3379 | |
| 3380 | // NB: The reason why we create converted objects in this convoluted way, |
| 3381 | // using a factory function instead of global variable, is that they |
| 3382 | // may be used at static initialization time (some of them are used by |
| 3383 | // wxString ctors and there may be a global wxString object). In other |
| 3384 | // words, possibly _before_ the converter global object would be |
| 3385 | // initialized. |
| 3386 | |
| 3387 | #undef wxConvLibc |
| 3388 | #undef wxConvUTF8 |
| 3389 | #undef wxConvUTF7 |
| 3390 | #undef wxConvLocal |
| 3391 | #undef wxConvISO8859_1 |
| 3392 | |
| 3393 | #define WX_DEFINE_GLOBAL_CONV2(klass, impl_klass, name, ctor_args) \ |
| 3394 | WXDLLIMPEXP_DATA_BASE(klass*) name##Ptr = NULL; \ |
| 3395 | WXDLLIMPEXP_BASE klass* wxGet_##name##Ptr() \ |
| 3396 | { \ |
| 3397 | static impl_klass name##Obj ctor_args; \ |
| 3398 | return &name##Obj; \ |
| 3399 | } \ |
| 3400 | /* this ensures that all global converter objects are created */ \ |
| 3401 | /* by the time static initialization is done, i.e. before any */ \ |
| 3402 | /* thread is launched: */ \ |
| 3403 | static klass* gs_##name##instance = wxGet_##name##Ptr() |
| 3404 | |
| 3405 | #define WX_DEFINE_GLOBAL_CONV(klass, name, ctor_args) \ |
| 3406 | WX_DEFINE_GLOBAL_CONV2(klass, klass, name, ctor_args) |
| 3407 | |
| 3408 | #ifdef __INTELC__ |
| 3409 | // disable warning "variable 'xxx' was declared but never referenced" |
| 3410 | #pragma warning(disable: 177) |
| 3411 | #endif // Intel C++ |
| 3412 | |
| 3413 | #ifdef __WINDOWS__ |
| 3414 | WX_DEFINE_GLOBAL_CONV2(wxMBConv, wxMBConv_win32, wxConvLibc, wxEMPTY_PARAMETER_VALUE); |
| 3415 | #elif 0 // defined(__WXOSX__) |
| 3416 | WX_DEFINE_GLOBAL_CONV2(wxMBConv, wxMBConv_cf, wxConvLibc, (wxFONTENCODING_UTF8)); |
| 3417 | #else |
| 3418 | WX_DEFINE_GLOBAL_CONV2(wxMBConv, wxMBConvLibc, wxConvLibc, wxEMPTY_PARAMETER_VALUE); |
| 3419 | #endif |
| 3420 | |
| 3421 | // NB: we can't use wxEMPTY_PARAMETER_VALUE as final argument here because it's |
| 3422 | // passed to WX_DEFINE_GLOBAL_CONV2 after a macro expansion and so still |
| 3423 | // provokes an error message about "not enough macro parameters"; and we |
| 3424 | // can't use "()" here as the name##Obj declaration would be parsed as a |
| 3425 | // function declaration then, so use a semicolon and live with an extra |
| 3426 | // empty statement (and hope that no compilers warns about this) |
| 3427 | WX_DEFINE_GLOBAL_CONV(wxMBConvStrictUTF8, wxConvUTF8, ;); |
| 3428 | WX_DEFINE_GLOBAL_CONV(wxMBConvUTF7, wxConvUTF7, ;); |
| 3429 | |
| 3430 | WX_DEFINE_GLOBAL_CONV(wxCSConv, wxConvLocal, (wxFONTENCODING_SYSTEM)); |
| 3431 | WX_DEFINE_GLOBAL_CONV(wxCSConv, wxConvISO8859_1, (wxFONTENCODING_ISO8859_1)); |
| 3432 | |
| 3433 | WXDLLIMPEXP_DATA_BASE(wxMBConv *) wxConvCurrent = wxGet_wxConvLibcPtr(); |
| 3434 | WXDLLIMPEXP_DATA_BASE(wxMBConv *) wxConvUI = wxGet_wxConvLocalPtr(); |
| 3435 | |
| 3436 | #ifdef __DARWIN__ |
| 3437 | // It is important to use this conversion object under Darwin as it ensures |
| 3438 | // that Unicode strings are (re)composed correctly even though xnu kernel uses |
| 3439 | // decomposed form internally (at least for the file names). |
| 3440 | static wxMBConv_cf wxConvMacUTF8DObj(wxFONTENCODING_UTF8); |
| 3441 | #endif |
| 3442 | |
| 3443 | WXDLLIMPEXP_DATA_BASE(wxMBConv *) wxConvFileName = |
| 3444 | #ifdef __DARWIN__ |
| 3445 | &wxConvMacUTF8DObj; |
| 3446 | #else // !__DARWIN__ |
| 3447 | wxGet_wxConvLibcPtr(); |
| 3448 | #endif // __DARWIN__/!__DARWIN__ |