标签:
终于搞完了,内容稍微有点多分为两篇把。
《指定存储文件的编码格式(上)》
《指定存储文件的编码格式(下)》
本篇为上篇。
主流的文件编码包括:UTF8\UTF8-WITHOUT-BOM\UTF16LE\UTF16BE\ANSI等。
中文的windows操作系统默认使用就是ANSI编码。
各种编码的主要规则大家可以去wiki网站、unicode.org网站等查看。
本文的上篇和下篇都采用windows函数WideCharToMultiByte和MultiByteToWideChar为基础进行编写的。
本文的上篇和下篇主要完成从指定文件A中读取数据,输出到指定编码的文件B中。
之所以分成上下两篇主要是下篇是对上篇的改进和优化。
本篇源码主要片段:
1 // test__OutputUtf8File.cpp : 定义控制台应用程序的入口点。 2 // 3 4 #include "stdafx.h" 5 #include <windows.h> 6 #include <string> 7 #include <iostream> 8 9 #define UTF8_SIGN 3 10 #define UTF16_SIGN 2 11 12 13 14 //************************************ 15 // Method: Utf16leFileToUtf8File 16 // FullName: Utf16leFileToUtf8File 17 // Access: public 18 // Returns: BOOL 19 // Qualifier:将lpUtf16leFile文件内容写入到lpUtf8File文件中 20 // Parameter: CONST LPTSTR lpUtf16leFile:输入文件utf16le编码 21 // Parameter: CONST LPTSTR lpUtf8File:输出文件为utf8-with-BOM编码 22 // *注:lpUtf16leFile文件只读;lpUtf8File文件总是创建或覆盖 23 //************************************ 24 BOOL Utf16leFileToUtf8File(CONST LPTSTR lpUtf16leFile, CONST LPTSTR lpUtf8File) 25 { 26 HANDLE hUtf16leFile = NULL; 27 HANDLE hUtf8File = NULL; 28 29 30 //create file 31 hUtf16leFile = ::CreateFile(lpUtf16leFile, GENERIC_READ, 0, NULL, OPEN_EXISTING, FILE_ATTRIBUTE_NORMAL, NULL); 32 if (INVALID_HANDLE_VALUE == hUtf16leFile) 33 { 34 int errLogNumber = GetLastError(); 35 printf_s("error number:%d\n", errLogNumber); 36 return FALSE; 37 } 38 39 40 //read UTF16LE encode file content 41 LPWSTR lpReadContentByUTF16 = NULL; 42 DWORD cbReadContentByUTF16 = 0; 43 DWORD cbPreReadContentByUTF16 = 0; 44 DWORD cchReadContentByUTF16 = 0; 45 46 cbReadContentByUTF16 = SetFilePointer(hUtf16leFile, 0, NULL, FILE_END); 47 if (INVALID_SET_FILE_POINTER == cbReadContentByUTF16) 48 { 49 int errLogNumber = GetLastError(); 50 printf_s("error number:%d\n", errLogNumber); 51 ::CloseHandle(hUtf16leFile); 52 return FALSE; 53 } 54 lpReadContentByUTF16 = (WCHAR *)malloc(cbReadContentByUTF16); 55 if (NULL == lpReadContentByUTF16) 56 { 57 printf_s("malloc error\n"); 58 ::CloseHandle(hUtf16leFile); 59 return FALSE; 60 } 61 ZeroMemory(lpReadContentByUTF16, cbReadContentByUTF16); 62 SetFilePointer(hUtf16leFile, 0, NULL, FILE_BEGIN); 63 if (FALSE == ReadFile(hUtf16leFile, lpReadContentByUTF16, cbReadContentByUTF16, &cbPreReadContentByUTF16, NULL)) 64 { 65 int errLogNumber = GetLastError(); 66 printf_s("error number:%d\n", errLogNumber); 67 free(lpReadContentByUTF16); 68 ::CloseHandle(hUtf16leFile); 69 return FALSE; 70 } 71 cchReadContentByUTF16 = ((cbReadContentByUTF16 % sizeof(WCHAR)) != 0 ? (cbReadContentByUTF16 / sizeof(WCHAR) + 1) : (cbReadContentByUTF16 / sizeof(WCHAR))); 72 73 74 //transform encode 75 LPSTR lpWriteContentByUTF8 = NULL; 76 DWORD cchWriteContentByUTF8 = 0; 77 DWORD cbWriteContentByUTF8 = 0; 78 DWORD cbPreWriteContentByUTF8 = 0; 79 80 cbWriteContentByUTF8 = ((cbReadContentByUTF16 % sizeof(WCHAR)) != 0 ? (cbReadContentByUTF16 + sizeof(WCHAR)) : (cbReadContentByUTF16)); 81 cchWriteContentByUTF8 = cchReadContentByUTF16; 82 lpWriteContentByUTF8 = (CHAR *)malloc(cbWriteContentByUTF8); 83 if (NULL == lpWriteContentByUTF8) 84 { 85 printf_s("malloc error\n"); 86 free(lpReadContentByUTF16); 87 ::CloseHandle(hUtf16leFile); 88 return FALSE; 89 } 90 ZeroMemory(lpWriteContentByUTF8, cbWriteContentByUTF8); 91 if (0 == WideCharToMultiByte(CP_UTF8, 0, lpReadContentByUTF16, cchReadContentByUTF16, lpWriteContentByUTF8, cbWriteContentByUTF8, NULL, NULL)) 92 { 93 printf_s("transform error\n"); 94 free(lpReadContentByUTF16); 95 free(lpWriteContentByUTF8); 96 ::CloseHandle(hUtf16leFile); 97 return FALSE; 98 } 99 100 101 //write UTF8 encode file content 102 hUtf8File = ::CreateFile(lpUtf8File, GENERIC_WRITE, 0, NULL, CREATE_ALWAYS, FILE_ATTRIBUTE_NORMAL, NULL); 103 if (INVALID_HANDLE_VALUE == hUtf8File) 104 { 105 printf("Terminal failure: Unable to write to file.\n"); 106 free(lpReadContentByUTF16); 107 free(lpWriteContentByUTF8); 108 ::CloseHandle(hUtf16leFile); 109 return FALSE; 110 } 111 for (int i = 0; i != cbWriteContentByUTF8; ++i) 112 { 113 if (TEXT(‘\0‘)==lpWriteContentByUTF8[i]) 114 { 115 cchWriteContentByUTF8 = i; 116 //lpWriteContentByUTF8[i] = 0;//设置结束符,但是utf8好像不用把 117 break; 118 } 119 } 120 SetFilePointer(hUtf8File, 0, NULL, FILE_BEGIN); 121 if (FALSE == WriteFile(hUtf8File, lpWriteContentByUTF8, cchWriteContentByUTF8, &cbPreWriteContentByUTF8, NULL)) 122 { 123 int errLogNumber = GetLastError(); 124 printf_s("error number:%d\n", errLogNumber); 125 free(lpReadContentByUTF16); 126 free(lpWriteContentByUTF8); 127 ::CloseHandle(hUtf16leFile); 128 return FALSE; 129 } 130 131 132 //release resource 133 free(lpReadContentByUTF16); 134 free(lpWriteContentByUTF8); 135 ::CloseHandle(hUtf16leFile); 136 ::CloseHandle(hUtf8File); 137 138 return TRUE; 139 140 } 141 142 //************************************ 143 // Method: Utf16leFileToUtf8NoBOMFile 144 // FullName: Utf16leFileToUtf8NoBOMFile 145 // Access: public 146 // Returns: BOOL 147 // Qualifier:将lpUtf16leFile文件内容写入到lpUtf8NoBOMFile文件中 148 // Parameter: CONST LPTSTR lpUtf16leFile:输入文件utf16le编码 149 // Parameter: CONST LPTSTR lpUtf8NoBOMFile:输出文件为utf8-without-BOM编码 150 // *注:lpUtf16leFile文件只读;lpUtf8NoBOMFile文件总是创建或覆盖 151 //************************************ 152 BOOL Utf16leFileToUtf8NoBOMFile(CONST LPTSTR lpUtf16leFile, CONST LPTSTR lpUtf8NoBOMFile) 153 { 154 HANDLE hUtf16leFile = NULL; 155 HANDLE hUtf8NoBOMFile = NULL; 156 157 158 //create UTF16LE file 159 hUtf16leFile = ::CreateFile(lpUtf16leFile, GENERIC_READ, 0, NULL, OPEN_EXISTING, FILE_ATTRIBUTE_NORMAL, NULL); 160 if (INVALID_HANDLE_VALUE == hUtf16leFile) 161 { 162 int errLogNumber = GetLastError(); 163 printf_s("error number:%d\n", errLogNumber); 164 return FALSE; 165 } 166 167 168 //read UTF16LE encode file content 169 LPWSTR lpReadContentByUTF16 = NULL; 170 DWORD cbReadContentByUTF16 = 0; 171 DWORD cbPreReadContentByUTF16 = 0; 172 DWORD cchReadContentByUTF16 = 0; 173 174 cbReadContentByUTF16 = SetFilePointer(hUtf16leFile, 0, NULL, FILE_END); 175 if (INVALID_SET_FILE_POINTER == cbReadContentByUTF16) 176 { 177 int errLogNumber = GetLastError(); 178 printf_s("error number:%d\n", errLogNumber); 179 ::CloseHandle(hUtf16leFile); 180 return FALSE; 181 } 182 lpReadContentByUTF16 = (WCHAR *)malloc(cbReadContentByUTF16); 183 if (NULL == lpReadContentByUTF16) 184 { 185 printf_s("malloc error\n"); 186 ::CloseHandle(hUtf16leFile); 187 return FALSE; 188 } 189 ZeroMemory(lpReadContentByUTF16, cbReadContentByUTF16); 190 SetFilePointer(hUtf16leFile, 0, NULL, FILE_BEGIN); 191 if (FALSE == ReadFile(hUtf16leFile, lpReadContentByUTF16, cbReadContentByUTF16, &cbPreReadContentByUTF16, NULL)) 192 { 193 int errLogNumber = GetLastError(); 194 printf_s("error number:%d\n", errLogNumber); 195 free(lpReadContentByUTF16); 196 ::CloseHandle(hUtf16leFile); 197 return FALSE; 198 } 199 cchReadContentByUTF16 = ((cbReadContentByUTF16 % sizeof(WCHAR)) != 0 ? (cbReadContentByUTF16 / sizeof(WCHAR) + 1) : (cbReadContentByUTF16 / sizeof(WCHAR))); 200 201 202 //transform encode 203 LPSTR lpWriteContentByUTF8 = NULL; 204 DWORD cchWriteContentByUTF8 = 0; 205 DWORD cbWriteContentByUTF8 = 0; 206 DWORD cbPreWriteContentByUTF8 = 0; 207 208 cbWriteContentByUTF8 = ((cbReadContentByUTF16 % sizeof(WCHAR)) != 0 ? (cbReadContentByUTF16 + sizeof(WCHAR)) : (cbReadContentByUTF16)); 209 cchWriteContentByUTF8 = cchReadContentByUTF16; 210 lpWriteContentByUTF8 = (CHAR *)malloc(cbWriteContentByUTF8); 211 if (NULL == lpWriteContentByUTF8) 212 { 213 printf_s("malloc error\n"); 214 free(lpReadContentByUTF16); 215 ::CloseHandle(hUtf16leFile); 216 return FALSE; 217 } 218 ZeroMemory(lpWriteContentByUTF8, cbWriteContentByUTF8); 219 if (0 == WideCharToMultiByte(CP_UTF8, 0, lpReadContentByUTF16, cchReadContentByUTF16, lpWriteContentByUTF8, cbWriteContentByUTF8, NULL, NULL)) 220 { 221 printf_s("transform error\n"); 222 free(lpReadContentByUTF16); 223 free(lpWriteContentByUTF8); 224 ::CloseHandle(hUtf16leFile); 225 return FALSE; 226 } 227 228 229 //write UTF8NoBOM encode file content 230 LPSTR lpWriteContentByUTF8NOBOM = NULL; 231 DWORD cbWriteContentByUTF8NOBOM = 0; 232 233 hUtf8NoBOMFile = ::CreateFile(lpUtf8NoBOMFile, GENERIC_WRITE, 0, NULL, CREATE_ALWAYS, FILE_ATTRIBUTE_NORMAL, NULL); 234 if (INVALID_HANDLE_VALUE == hUtf8NoBOMFile) 235 { 236 printf("Terminal failure: Unable to write to file.\n"); 237 free(lpReadContentByUTF16); 238 free(lpWriteContentByUTF8); 239 ::CloseHandle(hUtf16leFile); 240 return FALSE; 241 } 242 for (int i = 0; i != cbWriteContentByUTF8; ++i) 243 { 244 if (TEXT(‘\0‘) == lpWriteContentByUTF8[i]) 245 { 246 cbWriteContentByUTF8NOBOM = i - UTF8_SIGN; 247 break; 248 } 249 } 250 lpWriteContentByUTF8NOBOM = (CHAR *)malloc(cbWriteContentByUTF8NOBOM); 251 if (NULL == lpWriteContentByUTF8NOBOM) 252 { 253 printf_s("malloc error\n"); 254 free(lpReadContentByUTF16); 255 free(lpWriteContentByUTF8); 256 ::CloseHandle(hUtf16leFile); 257 return FALSE; 258 } 259 ZeroMemory(lpWriteContentByUTF8NOBOM, cbWriteContentByUTF8NOBOM); 260 CopyMemory(lpWriteContentByUTF8NOBOM, lpWriteContentByUTF8 + UTF8_SIGN, cbWriteContentByUTF8NOBOM); 261 SetFilePointer(hUtf8NoBOMFile, 0, NULL, FILE_BEGIN); 262 if (FALSE == WriteFile(hUtf8NoBOMFile, lpWriteContentByUTF8NOBOM, cbWriteContentByUTF8NOBOM, &cbPreWriteContentByUTF8, NULL)) 263 { 264 int errLogNumber = GetLastError(); 265 printf_s("error number:%d\n", errLogNumber); 266 free(lpReadContentByUTF16); 267 free(lpWriteContentByUTF8); 268 free(lpWriteContentByUTF8NOBOM); 269 ::CloseHandle(hUtf16leFile); 270 return FALSE; 271 } 272 273 274 //release resource 275 free(lpReadContentByUTF16); 276 free(lpWriteContentByUTF8); 277 free(lpWriteContentByUTF8NOBOM); 278 ::CloseHandle(hUtf16leFile); 279 ::CloseHandle(hUtf8NoBOMFile); 280 281 return TRUE; 282 283 } 284 285 //************************************ 286 // Method: Utf16leFileToUtf8NoBOMFile2 287 // FullName: Utf16leFileToUtf8NoBOMFile2 288 // Access: public 289 // Returns: BOOL 290 // Qualifier: 291 // Parameter: CONST LPTSTR lpUtf16leFile:输入文件utf16le编码 292 // Parameter: CONST LPTSTR lpUtf8NoBOMFile:输出文件为utf8-without-BOM编码 293 // *注:lpUtf16leFile文件只读;lpUtf8NoBOMFile文件总是创建或覆盖 294 //************************************ 295 BOOL Utf16leFileToUtf8NoBOMFile2(CONST LPTSTR lpUtf16leFile, CONST LPTSTR lpUtf8NoBOMFile) 296 { 297 HANDLE hUtf16leFile = NULL; 298 HANDLE hUtf8NoBOMFile = NULL; 299 300 301 //create UTF16LE file 302 hUtf16leFile = ::CreateFile(lpUtf16leFile, GENERIC_READ, 0, NULL, OPEN_EXISTING, FILE_ATTRIBUTE_NORMAL, NULL); 303 if (INVALID_HANDLE_VALUE == hUtf16leFile) 304 { 305 int errLogNumber = GetLastError(); 306 printf_s("error number:%d\n", errLogNumber); 307 return FALSE; 308 } 309 310 311 //read UTF16LE encode file content 312 LPWSTR lpReadContentByUTF16 = NULL; 313 DWORD cbReadContentByUTF16 = 0; 314 DWORD cbPreReadContentByUTF16 = 0; 315 DWORD cchReadContentByUTF16 = 0; 316 317 cbReadContentByUTF16 = SetFilePointer(hUtf16leFile, 0, NULL, FILE_END); 318 if (INVALID_SET_FILE_POINTER == cbReadContentByUTF16) 319 { 320 int errLogNumber = GetLastError(); 321 printf_s("error number:%d\n", errLogNumber); 322 ::CloseHandle(hUtf16leFile); 323 return FALSE; 324 } 325 lpReadContentByUTF16 = (WCHAR *)malloc(cbReadContentByUTF16); 326 if (NULL == lpReadContentByUTF16) 327 { 328 printf_s("malloc error\n"); 329 ::CloseHandle(hUtf16leFile); 330 return FALSE; 331 } 332 ZeroMemory(lpReadContentByUTF16, cbReadContentByUTF16); 333 SetFilePointer(hUtf16leFile, 0, NULL, FILE_BEGIN); 334 if (FALSE == ReadFile(hUtf16leFile, lpReadContentByUTF16, cbReadContentByUTF16, &cbPreReadContentByUTF16, NULL)) 335 { 336 int errLogNumber = GetLastError(); 337 printf_s("error number:%d\n", errLogNumber); 338 free(lpReadContentByUTF16); 339 ::CloseHandle(hUtf16leFile); 340 return FALSE; 341 } 342 cchReadContentByUTF16 = ((cbReadContentByUTF16 % sizeof(WCHAR)) != 0 ? (cbReadContentByUTF16 / sizeof(WCHAR) + 1) : (cbReadContentByUTF16 / sizeof(WCHAR))); 343 344 345 //transform encode 346 LPSTR lpWriteContentByUTF8 = NULL; 347 DWORD cchWriteContentByUTF8 = 0; 348 DWORD cbWriteContentByUTF8 = 0; 349 DWORD cbPreWriteContentByUTF8 = 0; 350 351 cbWriteContentByUTF8 = ((cbReadContentByUTF16 % sizeof(WCHAR)) != 0 ? (cbReadContentByUTF16 + sizeof(WCHAR)) : (cbReadContentByUTF16)); 352 cchWriteContentByUTF8 = cchReadContentByUTF16; 353 lpWriteContentByUTF8 = (CHAR *)malloc(cbWriteContentByUTF8); 354 if (NULL == lpWriteContentByUTF8) 355 { 356 printf_s("malloc error\n"); 357 free(lpReadContentByUTF16); 358 ::CloseHandle(hUtf16leFile); 359 return FALSE; 360 } 361 ZeroMemory(lpWriteContentByUTF8, cbWriteContentByUTF8); 362 if (0 == WideCharToMultiByte(CP_UTF8, 0, lpReadContentByUTF16, cchReadContentByUTF16, lpWriteContentByUTF8, cbWriteContentByUTF8, NULL, NULL)) 363 { 364 printf_s("transform error\n"); 365 free(lpReadContentByUTF16); 366 free(lpWriteContentByUTF8); 367 ::CloseHandle(hUtf16leFile); 368 return FALSE; 369 } 370 371 372 //write UTF8NOBOM file content 373 hUtf8NoBOMFile = ::CreateFile(lpUtf8NoBOMFile, GENERIC_WRITE, 0, NULL, CREATE_ALWAYS, FILE_ATTRIBUTE_NORMAL, NULL); 374 if (INVALID_HANDLE_VALUE == hUtf8NoBOMFile) 375 { 376 printf("Terminal failure: Unable to write to file.\n"); 377 free(lpReadContentByUTF16); 378 free(lpWriteContentByUTF8); 379 ::CloseHandle(hUtf16leFile); 380 return FALSE; 381 } 382 for (int i = 0; i != cbWriteContentByUTF8; ++i) 383 { 384 if (TEXT(‘\0‘) == lpWriteContentByUTF8[i]) 385 { 386 cchWriteContentByUTF8 = i; 387 break; 388 } 389 } 390 SetFilePointer(hUtf8NoBOMFile, 0, NULL, FILE_BEGIN); 391 if (FALSE == WriteFile(hUtf8NoBOMFile, lpWriteContentByUTF8 + UTF8_SIGN, cchWriteContentByUTF8 - UTF8_SIGN, &cbPreWriteContentByUTF8, NULL)) 392 { 393 int errLogNumber = GetLastError(); 394 printf_s("error number:%d\n", errLogNumber); 395 free(lpReadContentByUTF16); 396 free(lpWriteContentByUTF8); 397 ::CloseHandle(hUtf16leFile); 398 ::CloseHandle(hUtf8NoBOMFile); 399 return FALSE; 400 } 401 402 403 //release resource 404 free(lpReadContentByUTF16); 405 free(lpWriteContentByUTF8); 406 ::CloseHandle(hUtf16leFile); 407 ::CloseHandle(hUtf8NoBOMFile); 408 409 return TRUE; 410 } 411 412 //************************************ 413 // Method: Utf8FileToUtf16leFile 414 // FullName: Utf8FileToUtf16leFile 415 // Access: public 416 // Returns: BOOL 417 // Qualifier:将utf8编码格式文件转换为utf16le编码格式文件 418 // Parameter: CONST LPTSTR lpUtf8File:utf8编码格式文件 419 // Parameter: CONST LPTSTR lpUtf16leFile:utf16le编码格式文件 420 //************************************ 421 BOOL Utf8FileToUtf16leFile(CONST LPTSTR lpUtf8File, CONST LPTSTR lpUtf16leFile) 422 { 423 HANDLE hUtf16leFile = NULL; 424 HANDLE hUtf8File = NULL; 425 426 427 //create UTF8 file 428 hUtf8File = ::CreateFile(lpUtf8File, GENERIC_READ, 0, NULL, OPEN_EXISTING, FILE_ATTRIBUTE_NORMAL, NULL); 429 if (INVALID_HANDLE_VALUE == hUtf8File) 430 { 431 int errLogNumber = GetLastError(); 432 printf_s("error number:%d\n", errLogNumber); 433 return FALSE; 434 } 435 436 437 //read UTF8 encode file content 438 LPSTR lpReadContentByUTF8 = NULL; 439 DWORD cbReadContentByUTF8 = 0; 440 DWORD cbPreReadContentByUTF8 = 0; 441 DWORD cchReadContentByUTF8 = 0; 442 443 cbReadContentByUTF8 = SetFilePointer(hUtf8File, 0, NULL, FILE_END); 444 if (INVALID_SET_FILE_POINTER == cbReadContentByUTF8) 445 { 446 int errLogNumber = GetLastError(); 447 printf_s("error number:%d\n", errLogNumber); 448 ::CloseHandle(hUtf8File); 449 return FALSE; 450 } 451 lpReadContentByUTF8 = (CHAR *)malloc(cbReadContentByUTF8); 452 if (NULL == lpReadContentByUTF8) 453 { 454 printf_s("malloc error\n"); 455 ::CloseHandle(hUtf8File); 456 return FALSE; 457 } 458 ZeroMemory(lpReadContentByUTF8, cbReadContentByUTF8); 459 SetFilePointer(hUtf8File, 0, NULL, FILE_BEGIN); 460 if (FALSE == ReadFile(hUtf8File, lpReadContentByUTF8, cbReadContentByUTF8, &cbPreReadContentByUTF8, NULL)) 461 { 462 int errLogNumber = GetLastError(); 463 printf_s("error number:%d\n", errLogNumber); 464 free(lpReadContentByUTF8); 465 ::CloseHandle(hUtf8File); 466 return FALSE; 467 } 468 if (sizeof(CHAR) != sizeof(BYTE)) 469 { 470 free(lpReadContentByUTF8); 471 ::CloseHandle(hUtf8File); 472 return FALSE; 473 } 474 cchReadContentByUTF8 = cbReadContentByUTF8; 475 476 477 //transform encode 478 LPWSTR lpWriteContentByUTF16 = NULL; 479 DWORD cchWriteContentByUTF16 = 0; 480 DWORD cbWriteContentByUTF16 = 0; 481 DWORD cbPreWriteContentByUTF16 = 0; 482 483 cbWriteContentByUTF16 = cchReadContentByUTF8 * 2; 484 cchWriteContentByUTF16 = cchReadContentByUTF8; 485 lpWriteContentByUTF16 = (WCHAR *)malloc(cbWriteContentByUTF16); 486 if (NULL == lpWriteContentByUTF16) 487 { 488 printf_s("malloc error\n"); 489 free(lpReadContentByUTF8); 490 ::CloseHandle(hUtf8File); 491 return FALSE; 492 } 493 ZeroMemory(lpWriteContentByUTF16, cbWriteContentByUTF16); 494 if (0 == MultiByteToWideChar(CP_UTF8, 0, lpReadContentByUTF8, cbReadContentByUTF8, lpWriteContentByUTF16, cchWriteContentByUTF16)) 495 { 496 printf_s("transform error\n"); 497 free(lpReadContentByUTF8); 498 free(lpWriteContentByUTF16); 499 ::CloseHandle(hUtf8File); 500 return FALSE; 501 } 502 503 504 //write UTF16LE encode file content 505 hUtf16leFile = ::CreateFile(lpUtf16leFile, GENERIC_WRITE, 0, NULL, CREATE_ALWAYS, FILE_ATTRIBUTE_NORMAL, NULL); 506 if (INVALID_HANDLE_VALUE == hUtf16leFile) 507 { 508 printf("Terminal failure: Unable to write to file.\n"); 509 free(lpReadContentByUTF8); 510 free(lpWriteContentByUTF16); 511 ::CloseHandle(hUtf8File); 512 return FALSE; 513 } 514 for (int i = 0; i != cbWriteContentByUTF16 -1; ++i) 515 { 516 if (0 == lpWriteContentByUTF16[i] && 0 == lpWriteContentByUTF16[i+1]) 517 { 518 cbWriteContentByUTF16 = i; 519 break; 520 } 521 } 522 ////////////////////////////////////////////////////////////////////////// 523 524 //std::wstring wstrText = L"output to utf8 with BOM.\r\n输出到utf8带BOM的文件中。\r\nengli sh"; 525 //int lgText = wstrText.length(); 526 //int lgText2 = wstrText.size(); 527 //std::cout << lgText << "" << lgText2 << std::endl; 528 529 ////////////////////////////////////////////////////////////////////////// 530 SetFilePointer(hUtf16leFile, 0, NULL, FILE_BEGIN); 531 if (FALSE == WriteFile(hUtf16leFile, lpWriteContentByUTF16, cbWriteContentByUTF16 * sizeof(WCHAR), &cbPreWriteContentByUTF16, NULL)) 532 { 533 int errLogNumber = GetLastError(); 534 printf_s("error number:%d\n", errLogNumber); 535 free(lpReadContentByUTF8); 536 free(lpWriteContentByUTF16); 537 ::CloseHandle(hUtf16leFile); 538 ::CloseHandle(hUtf8File); 539 return FALSE; 540 } 541 542 543 //release resource 544 free(lpReadContentByUTF8); 545 free(lpWriteContentByUTF16); 546 ::CloseHandle(hUtf16leFile); 547 ::CloseHandle(hUtf8File); 548 549 return TRUE; 550 551 } 552 553 BOOL Utf8NoBOMFileToUtf16leFile(CONST LPTSTR lpUtfNoBOM8File, CONST LPTSTR lpUtf16leFile) 554 { 555 return Utf8FileToUtf16leFile(lpUtfNoBOM8File, lpUtf16leFile); 556 557 //HANDLE hUtf16leFile = NULL; 558 //HANDLE hUtf8File = NULL; 559 560 561 ////create UTF8 file 562 //hUtf8File = ::CreateFile(lpUtf8File, GENERIC_READ, 0, NULL, OPEN_EXISTING, FILE_ATTRIBUTE_NORMAL, NULL); 563 //if (INVALID_HANDLE_VALUE == hUtf8File) 564 //{ 565 // int errLogNumber = GetLastError(); 566 // printf_s("error number:%d\n", errLogNumber); 567 // return FALSE; 568 //} 569 570 571 ////read UTF8 encode file content 572 //LPSTR lpReadContentByUTF8 = NULL; 573 //DWORD cbReadContentByUTF8 = 0; 574 //DWORD cbPreReadContentByUTF8 = 0; 575 //DWORD cchReadContentByUTF8 = 0; 576 577 //cbReadContentByUTF8 = SetFilePointer(hUtf8File, 0, NULL, FILE_END); 578 //if (INVALID_SET_FILE_POINTER == cbReadContentByUTF8) 579 //{ 580 // int errLogNumber = GetLastError(); 581 // printf_s("error number:%d\n", errLogNumber); 582 // ::CloseHandle(hUtf8File); 583 // return FALSE; 584 //} 585 //lpReadContentByUTF8 = (CHAR *)malloc(cbReadContentByUTF8); 586 //if (NULL == lpReadContentByUTF8) 587 //{ 588 // printf_s("malloc error\n"); 589 // ::CloseHandle(hUtf8File); 590 // return FALSE; 591 //} 592 //ZeroMemory(lpReadContentByUTF8, cbReadContentByUTF8); 593 //SetFilePointer(hUtf8File, 0, NULL, FILE_BEGIN); 594 //if (FALSE == ReadFile(hUtf8File, lpReadContentByUTF8, cbReadContentByUTF8, &cbPreReadContentByUTF8, NULL)) 595 //{ 596 // int errLogNumber = GetLastError(); 597 // printf_s("error number:%d\n", errLogNumber); 598 // free(lpReadContentByUTF8); 599 // ::CloseHandle(hUtf8File); 600 // return FALSE; 601 //} 602 //if (sizeof(CHAR) != sizeof(BYTE)) 603 //{ 604 // free(lpReadContentByUTF8); 605 // ::CloseHandle(hUtf8File); 606 // return FALSE; 607 //} 608 //cchReadContentByUTF8 = cbReadContentByUTF8; 609 610 611 ////transform encode 612 //LPWSTR lpWriteContentByUTF16 = NULL; 613 //DWORD cchWriteContentByUTF16 = 0; 614 //DWORD cbWriteContentByUTF16 = 0; 615 //DWORD cbPreWriteContentByUTF16 = 0; 616 617 //cbWriteContentByUTF16 = cchReadContentByUTF8 * 2; 618 //cchWriteContentByUTF16 = cchReadContentByUTF8; 619 //lpWriteContentByUTF16 = (WCHAR *)malloc(cbWriteContentByUTF16); 620 //if (NULL == lpWriteContentByUTF16) 621 //{ 622 // printf_s("malloc error\n"); 623 // free(lpReadContentByUTF8); 624 // ::CloseHandle(hUtf8File); 625 // return FALSE; 626 //} 627 //ZeroMemory(lpWriteContentByUTF16, cbWriteContentByUTF16); 628 //if (0 == MultiByteToWideChar(CP_UTF8, 0, lpReadContentByUTF8, cbReadContentByUTF8, lpWriteContentByUTF16, cchWriteContentByUTF16)) 629 //{ 630 // printf_s("transform error\n"); 631 // free(lpReadContentByUTF8); 632 // free(lpWriteContentByUTF16); 633 // ::CloseHandle(hUtf8File); 634 // return FALSE; 635 //} 636 637 638 ////write UTF16LE encode file content 639 //hUtf16leFile = ::CreateFile(lpUtf16leFile, GENERIC_WRITE, 0, NULL, CREATE_ALWAYS, FILE_ATTRIBUTE_NORMAL, NULL); 640 //if (INVALID_HANDLE_VALUE == hUtf16leFile) 641 //{ 642 // printf("Terminal failure: Unable to write to file.\n"); 643 // free(lpReadContentByUTF8); 644 // free(lpWriteContentByUTF16); 645 // ::CloseHandle(hUtf8File); 646 // return FALSE; 647 //} 648 //for (int i = 0; i != cbWriteContentByUTF16 - 1; ++i) 649 //{ 650 // if (0 == lpWriteContentByUTF16[i] && 0 == lpWriteContentByUTF16[i + 1]) 651 // { 652 // cbWriteContentByUTF16 = i; 653 // break; 654 // } 655 //} 656 //////////////////////////////////////////////////////////////////////////// 657 658 ////std::wstring wstrText = L"output to utf8 with BOM.\r\n输出到utf8带BOM的文件中。\r\nengli sh"; 659 ////int lgText = wstrText.length(); 660 ////int lgText2 = wstrText.size(); 661 ////std::cout << lgText << "" << lgText2 << std::endl; 662 663 //////////////////////////////////////////////////////////////////////////// 664 //SetFilePointer(hUtf16leFile, 0, NULL, FILE_BEGIN); 665 //if (FALSE == WriteFile(hUtf16leFile, lpWriteContentByUTF16, cbWriteContentByUTF16 * sizeof(WCHAR), &cbPreWriteContentByUTF16, NULL)) 666 //{ 667 // int errLogNumber = GetLastError(); 668 // printf_s("error number:%d\n", errLogNumber); 669 // free(lpReadContentByUTF8); 670 // free(lpWriteContentByUTF16); 671 // ::CloseHandle(hUtf16leFile); 672 // ::CloseHandle(hUtf8File); 673 // return FALSE; 674 //} 675 676 677 ////release resource 678 //free(lpReadContentByUTF8); 679 //free(lpWriteContentByUTF16); 680 //::CloseHandle(hUtf16leFile); 681 //::CloseHandle(hUtf8File); 682 683 //return TRUE; 684 685 } 686 687 BOOL Utf8FileToUtf8NoBOMFile(CONST LPTSTR lpUtf8File, CONST LPTSTR lpUtf8NoBOMFile) 688 { 689 HANDLE hUtf8NoBOMFile = NULL; 690 HANDLE hUtf8File = NULL; 691 692 693 //create file 694 hUtf8File = ::CreateFile(lpUtf8File, GENERIC_READ, 0, NULL, OPEN_EXISTING, FILE_ATTRIBUTE_NORMAL, NULL); 695 if (INVALID_HANDLE_VALUE == hUtf8File) 696 { 697 int errLogNumber = GetLastError(); 698 printf_s("error number:%d\n", errLogNumber); 699 return FALSE; 700 } 701 702 703 //read UTF8 encode file content 704 LPSTR lpReadContentByUTF8 = NULL; 705 DWORD cbReadContentByUTF8 = 0; 706 DWORD cbPreReadContentByUTF8 = 0; 707 DWORD cchReadContentByUTF8 = 0; 708 709 cbReadContentByUTF8 = SetFilePointer(hUtf8File, 0, NULL, FILE_END); 710 if (INVALID_SET_FILE_POINTER == cbReadContentByUTF8) 711 { 712 int errLogNumber = GetLastError(); 713 printf_s("error number:%d\n", errLogNumber); 714 ::CloseHandle(hUtf8File); 715 return FALSE; 716 } 717 lpReadContentByUTF8 = (CHAR *)malloc(cbReadContentByUTF8); 718 if (NULL == lpReadContentByUTF8) 719 { 720 printf_s("malloc error\n"); 721 ::CloseHandle(hUtf8File); 722 return FALSE; 723 } 724 ZeroMemory(lpReadContentByUTF8, cbReadContentByUTF8); 725 SetFilePointer(hUtf8File, 0, NULL, FILE_BEGIN); 726 if (FALSE == ReadFile(hUtf8File, lpReadContentByUTF8, cbReadContentByUTF8, &cbPreReadContentByUTF8, NULL)) 727 { 728 int errLogNumber = GetLastError(); 729 printf_s("error number:%d\n", errLogNumber); 730 free(lpReadContentByUTF8); 731 ::CloseHandle(hUtf8File); 732 return FALSE; 733 } 734 if (sizeof(CHAR) != sizeof(BYTE)) 735 { 736 free(lpReadContentByUTF8); 737 ::CloseHandle(hUtf8File); 738 return FALSE; 739 } 740 cchReadContentByUTF8 = cbReadContentByUTF8; 741 742 743 //write UTF8NoBOM encode file content 744 LPSTR lpWriteContentByUTF8NoBOM = NULL; 745 DWORD cbWriteContentByUTF8NoBOM = 0; 746 DWORD cbPreWriteContentByUTF8NoBOM = 0; 747 DWORD cchWriteContentByUTF8NoBOM = 0; 748 749 cbWriteContentByUTF8NoBOM = cbReadContentByUTF8 - UTF8_SIGN; 750 lpWriteContentByUTF8NoBOM = (CHAR *)malloc(cbWriteContentByUTF8NoBOM); 751 if (NULL == lpWriteContentByUTF8NoBOM) 752 { 753 printf_s("malloc error\n"); 754 free(lpReadContentByUTF8); 755 ::CloseHandle(hUtf8File); 756 return FALSE; 757 } 758 ZeroMemory(lpWriteContentByUTF8NoBOM, cbWriteContentByUTF8NoBOM); 759 760 CopyMemory(lpWriteContentByUTF8NoBOM, lpReadContentByUTF8 + UTF8_SIGN, cbWriteContentByUTF8NoBOM); 761 762 hUtf8NoBOMFile = ::CreateFile(lpUtf8NoBOMFile, GENERIC_WRITE, 0, NULL, CREATE_ALWAYS, FILE_ATTRIBUTE_NORMAL, NULL); 763 if (INVALID_HANDLE_VALUE == hUtf8NoBOMFile) 764 { 765 printf("Terminal failure: Unable to write to file.\n"); 766 free(lpReadContentByUTF8); 767 free(lpWriteContentByUTF8NoBOM); 768 ::CloseHandle(hUtf8File); 769 return FALSE; 770 } 771 SetFilePointer(hUtf8NoBOMFile, 0, NULL, FILE_BEGIN); 772 if (FALSE == WriteFile(hUtf8NoBOMFile, lpWriteContentByUTF8NoBOM, cbWriteContentByUTF8NoBOM, &cbPreWriteContentByUTF8NoBOM, NULL)) 773 { 774 int errLogNumber = GetLastError(); 775 printf_s("error number:%d\n", errLogNumber); 776 free(lpReadContentByUTF8); 777 free(lpWriteContentByUTF8NoBOM); 778 ::CloseHandle(hUtf8NoBOMFile); 779 ::CloseHandle(hUtf8File); 780 return FALSE; 781 } 782 783 784 //release resource 785 free(lpReadContentByUTF8); 786 free(lpWriteContentByUTF8NoBOM); 787 ::CloseHandle(hUtf8NoBOMFile); 788 ::CloseHandle(hUtf8File); 789 790 return TRUE; 791 792 } 793 794 BOOL Utf8NoBOMFileToUtf8File(CONST LPTSTR lpUtf8NoBOMFile, CONST LPTSTR lpUtf8File) 795 { 796 HANDLE hUtf8NoBOMFile = NULL; 797 HANDLE hUtf8File = NULL; 798 799 800 //create file 801 hUtf8NoBOMFile = ::CreateFile(lpUtf8NoBOMFile, GENERIC_READ, 0, NULL, OPEN_EXISTING, FILE_ATTRIBUTE_NORMAL, NULL); 802 if (INVALID_HANDLE_VALUE == hUtf8NoBOMFile) 803 { 804 int errLogNumber = GetLastError(); 805 printf_s("error number:%d\n", errLogNumber); 806 return FALSE; 807 } 808 809 810 //read UTF8 encode file content 811 LPSTR lpReadContentByUTF8NoBOM = NULL; 812 DWORD cbReadContentByUTF8NoBOM = 0; 813 DWORD cbPreReadContentByUTF8NoBOM = 0; 814 DWORD cchReadContentByUTF8NoBOM = 0; 815 816 cbReadContentByUTF8NoBOM = SetFilePointer(hUtf8NoBOMFile, 0, NULL, FILE_END); 817 if (INVALID_SET_FILE_POINTER == cbReadContentByUTF8NoBOM) 818 { 819 int errLogNumber = GetLastError(); 820 printf_s("error number:%d\n", errLogNumber); 821 ::CloseHandle(hUtf8NoBOMFile); 822 return FALSE; 823 } 824 lpReadContentByUTF8NoBOM = (CHAR *)malloc(cbReadContentByUTF8NoBOM); 825 if (NULL == lpReadContentByUTF8NoBOM) 826 { 827 printf_s("malloc error\n"); 828 ::CloseHandle(hUtf8NoBOMFile); 829 return FALSE; 830 } 831 ZeroMemory(lpReadContentByUTF8NoBOM, cbReadContentByUTF8NoBOM); 832 SetFilePointer(hUtf8NoBOMFile, 0, NULL, FILE_BEGIN); 833 if (FALSE == ReadFile(hUtf8NoBOMFile, lpReadContentByUTF8NoBOM, cbReadContentByUTF8NoBOM, &cbPreReadContentByUTF8NoBOM, NULL)) 834 { 835 int errLogNumber = GetLastError(); 836 printf_s("error number:%d\n", errLogNumber); 837 free(lpReadContentByUTF8NoBOM); 838 ::CloseHandle(hUtf8NoBOMFile); 839 return FALSE; 840 } 841 if (sizeof(CHAR) != sizeof(BYTE)) 842 { 843 free(lpReadContentByUTF8NoBOM); 844 ::CloseHandle(hUtf8NoBOMFile); 845 return FALSE; 846 } 847 cchReadContentByUTF8NoBOM = cbReadContentByUTF8NoBOM; 848 849 850 //write UTF8NoBOM encode file content 851 LPSTR lpWriteContentByUTF8 = NULL; 852 DWORD cbWriteContentByUTF8 = 0; 853 DWORD cbPreWriteContentByUTF8 = 0; 854 DWORD cchWriteContentByUTF8 = 0; 855 856 cbWriteContentByUTF8 = cbReadContentByUTF8NoBOM + UTF8_SIGN; 857 lpWriteContentByUTF8 = (CHAR *)malloc(cbWriteContentByUTF8); 858 if (NULL == lpWriteContentByUTF8) 859 { 860 printf_s("malloc error\n"); 861 free(lpReadContentByUTF8NoBOM); 862 ::CloseHandle(hUtf8NoBOMFile); 863 return FALSE; 864 } 865 ZeroMemory(lpWriteContentByUTF8, cbWriteContentByUTF8); 866 867 lpWriteContentByUTF8[0] = 0xef; 868 lpWriteContentByUTF8[1] = 0xbb; 869 lpWriteContentByUTF8[2] = 0xbf; 870 CopyMemory(lpWriteContentByUTF8 + UTF8_SIGN, lpReadContentByUTF8NoBOM, cbWriteContentByUTF8 - UTF8_SIGN); 871 872 hUtf8File = ::CreateFile(lpUtf8File, GENERIC_WRITE, 0, NULL, CREATE_ALWAYS, FILE_ATTRIBUTE_NORMAL, NULL); 873 if (INVALID_HANDLE_VALUE == hUtf8File) 874 { 875 printf("Terminal failure: Unable to write to file.\n"); 876 free(lpWriteContentByUTF8); 877 free(lpReadContentByUTF8NoBOM); 878 ::CloseHandle(hUtf8NoBOMFile); 879 return FALSE; 880 } 881 SetFilePointer(hUtf8File, 0, NULL, FILE_BEGIN); 882 if (FALSE == WriteFile(hUtf8File, lpWriteContentByUTF8, cbWriteContentByUTF8, &cbPreWriteContentByUTF8, NULL)) 883 { 884 int errLogNumber = GetLastError(); 885 printf_s("error number:%d\n", errLogNumber); 886 free(lpWriteContentByUTF8); 887 free(lpReadContentByUTF8NoBOM); 888 ::CloseHandle(hUtf8NoBOMFile); 889 ::CloseHandle(hUtf8File); 890 return FALSE; 891 } 892 893 894 //release resource 895 free(lpWriteContentByUTF8); 896 free(lpReadContentByUTF8NoBOM); 897 ::CloseHandle(hUtf8NoBOMFile); 898 ::CloseHandle(hUtf8File); 899 900 return TRUE; 901 902 } 903 904 BOOL Utf16leFileToUtf16beFile(CONST LPTSTR lpUtf16leFile, CONST LPTSTR lpUtf16beFile) 905 { 906 HANDLE hUtf16leFile = NULL; 907 HANDLE hUtf16beFile = NULL; 908 909 910 //create file 911 hUtf16leFile = ::CreateFile(lpUtf16leFile, GENERIC_READ, 0, NULL, OPEN_EXISTING, FILE_ATTRIBUTE_NORMAL, NULL); 912 if (INVALID_HANDLE_VALUE == hUtf16leFile) 913 { 914 int errLogNumber = GetLastError(); 915 printf_s("error number:%d\n", errLogNumber); 916 return FALSE; 917 } 918 919 920 //read UTF16le encode file content 921 LPBYTE lpReadContentByUTF16le = NULL; 922 DWORD cbReadContentByUTF16le = 0; 923 DWORD cbPreReadContentByUTF16le = 0; 924 DWORD cchReadContentByUTF16le = 0; 925 926 cbReadContentByUTF16le = SetFilePointer(hUtf16leFile, 0, NULL, FILE_END); 927 if (INVALID_SET_FILE_POINTER == cbReadContentByUTF16le) 928 { 929 int errLogNumber = GetLastError(); 930 printf_s("error number:%d\n", errLogNumber); 931 ::CloseHandle(hUtf16leFile); 932 return FALSE; 933 } 934 if (0 != cbReadContentByUTF16le % 2) 935 { 936 printf_s("read byte error\n"); 937 ::CloseHandle(hUtf16leFile); 938 return FALSE; 939 } 940 lpReadContentByUTF16le = (BYTE *)malloc(cbReadContentByUTF16le); 941 if (NULL == lpReadContentByUTF16le) 942 { 943 printf_s("malloc error\n"); 944 ::CloseHandle(hUtf16leFile); 945 return FALSE; 946 } 947 ZeroMemory(lpReadContentByUTF16le, cbReadContentByUTF16le); 948 SetFilePointer(hUtf16leFile, 0, NULL, FILE_BEGIN); 949 if (FALSE == ReadFile(hUtf16leFile, lpReadContentByUTF16le, cbReadContentByUTF16le, &cbPreReadContentByUTF16le, NULL)) 950 { 951 int errLogNumber = GetLastError(); 952 printf_s("error number:%d\n", errLogNumber); 953 free(lpReadContentByUTF16le); 954 ::CloseHandle(hUtf16leFile); 955 return FALSE; 956 } 957 //std::wstring wstrText(lpReadContentByUTF16le); 958 //printf("%s\n", wstrText); 959 if (sizeof(WCHAR) != sizeof(BYTE)*2) 960 { 961 free(lpReadContentByUTF16le); 962 ::CloseHandle(hUtf16leFile); 963 return FALSE; 964 } 965 966 967 //write UTF8NoBOM encode file content 968 LPBYTE lpWriteContentByUTF16be = NULL; 969 DWORD cbWriteContentByUTF16be = 0; 970 DWORD cbPreWriteContentByUTF16be = 0; 971 DWORD cchWriteContentByUTF16be = 0; 972 973 cbWriteContentByUTF16be = cbReadContentByUTF16le; 974 lpWriteContentByUTF16be = (BYTE *)malloc(cbWriteContentByUTF16be); 975 if (NULL == lpWriteContentByUTF16be) 976 { 977 printf_s("malloc error\n"); 978 free(lpReadContentByUTF16le); 979 ::CloseHandle(hUtf16leFile); 980 return FALSE; 981 } 982 ZeroMemory(lpWriteContentByUTF16be, cbWriteContentByUTF16be); 983 984 CopyMemory(lpWriteContentByUTF16be, lpReadContentByUTF16le, cbWriteContentByUTF16be); 985 986 for (DWORD i = 0; i < cbWriteContentByUTF16be; i += 2)//每两值交换 987 { 988 lpWriteContentByUTF16be[i] = lpWriteContentByUTF16be[i] ^ lpWriteContentByUTF16be[i + 1]; 989 lpWriteContentByUTF16be[i + 1] = lpWriteContentByUTF16be[i + 1] ^ lpWriteContentByUTF16be[i]; 990 lpWriteContentByUTF16be[i] = lpWriteContentByUTF16be[i] ^ lpWriteContentByUTF16be[i + 1]; 991 //BYTE hex_ = 0x0; 992 //hex_ = lpWriteContentByUTF16be[i]; 993 ////printf("%x\n", lpWriteContentByUTF16be[i]); 994 //lpWriteContentByUTF16be[i] = lpWriteContentByUTF16be[i + 1]; 995 //lpWriteContentByUTF16be[i + 1] = hex_; 996 } 997 998 hUtf16beFile = ::CreateFile(lpUtf16beFile, GENERIC_WRITE, 0, NULL, CREATE_ALWAYS, FILE_ATTRIBUTE_NORMAL, NULL); 999 if (INVALID_HANDLE_VALUE == hUtf16beFile) 1000 { 1001 printf("Terminal failure: Unable to write to file.\n"); 1002 free(lpWriteContentByUTF16be); 1003 free(lpReadContentByUTF16le); 1004 ::CloseHandle(hUtf16leFile); 1005 return FALSE; 1006 } 1007 SetFilePointer(hUtf16beFile, 0, NULL, FILE_BEGIN); 1008 if (FALSE == WriteFile(hUtf16beFile, lpWriteContentByUTF16be, cbWriteContentByUTF16be, &cbPreWriteContentByUTF16be, NULL)) 1009 { 1010 int errLogNumber = GetLastError(); 1011 printf_s("error number:%d\n", errLogNumber); 1012 free(lpWriteContentByUTF16be); 1013 free(lpReadContentByUTF16le); 1014 ::CloseHandle(hUtf16leFile); 1015 ::CloseHandle(hUtf16beFile); 1016 return FALSE; 1017 } 1018 1019 1020 //release resource 1021 free(lpWriteContentByUTF16be); 1022 free(lpReadContentByUTF16le); 1023 ::CloseHandle(hUtf16leFile); 1024 ::CloseHandle(hUtf16beFile); 1025 1026 return TRUE; 1027 1028 } 1029 1030 BOOL Utf16beFileToUtf16leFile(CONST LPTSTR lpUtf16beFile, CONST LPTSTR lpUtf16leFile) 1031 { 1032 return Utf16leFileToUtf16beFile(lpUtf16beFile, lpUtf16leFile); 1033 } 1034 1035 1036 1037 1038 enum FileCodeType 1039 { 1040 OTHER = 0, 1041 UTF16LE, 1042 UTF8, 1043 UTF8_NO_BOM, 1044 UTF16BE 1045 }; 1046 1047 //************************************ 1048 // Method: CodeFileAToCodeFileB 1049 // FullName: CodeFileAToCodeFileB 1050 // Access: public 1051 // Returns: bool 1052 // Qualifier:读取指定编码文件A的内容输出到指定编码文件B 1053 // Parameter: CONST LPTSTR lpFileA:输入文件 1054 // Parameter: CONST FileCodeType wCodeTypeA:输入文件类型(包括:utf16le\utf16be\utf8\ANSI) 1055 // Parameter: CONST LPTSTR lpFileB:输出文件 1056 // Parameter: CONST FileCodeType wCodeTypeB:输出文件类型(包括:utf16le\utf16be\utf8\utf8 without BOM\ANSI) 1057 //************************************ 1058 BOOL CodeFileAToCodeFileB(CONST LPTSTR lpFileA, CONST FileCodeType emCodeTypeA, CONST LPTSTR lpFileB, CONST FileCodeType emCodeTypeB) 1059 { 1060 BOOL bSUCCESS = FALSE; 1061 HANDLE hFileA = NULL; 1062 HANDLE hFileB = NULL; 1063 1064 if (OTHER == emCodeTypeA || OTHER == emCodeTypeB) 1065 { 1066 return bSUCCESS; 1067 } 1068 if (NULL == lpFileA || NULL == lpFileB) 1069 { 1070 return bSUCCESS; 1071 } 1072 1073 switch (emCodeTypeA) 1074 { 1075 case UTF16LE: 1076 { 1077 switch (emCodeTypeB) 1078 { 1079 case UTF16BE: 1080 { 1081 return Utf16leFileToUtf16beFile(lpFileA, lpFileB); 1082 } 1083 break; 1084 case UTF8: 1085 { 1086 return Utf16leFileToUtf8File(lpFileA, lpFileB); 1087 } 1088 break; 1089 case UTF8_NO_BOM: 1090 { 1091 return Utf16leFileToUtf8NoBOMFile2(lpFileA, lpFileB); 1092 } 1093 break; 1094 default:; 1095 } 1096 } 1097 break; 1098 case UTF8: 1099 { 1100 switch (emCodeTypeB) 1101 { 1102 case UTF16LE: 1103 { 1104 return Utf8FileToUtf16leFile(lpFileA, lpFileB); 1105 } 1106 break; 1107 case UTF16BE: 1108 { 1109 1110 } 1111 break; 1112 case UTF8_NO_BOM: 1113 { 1114 return Utf8FileToUtf8NoBOMFile(lpFileA, lpFileB); 1115 } 1116 break; 1117 default:; 1118 } 1119 } 1120 break; 1121 case UTF8_NO_BOM: 1122 { 1123 switch (emCodeTypeB) 1124 { 1125 case UTF16LE: 1126 { 1127 return Utf8NoBOMFileToUtf16leFile(lpFileA, lpFileB); 1128 } 1129 break; 1130 case UTF8: 1131 { 1132 return Utf8NoBOMFileToUtf8File(lpFileA, lpFileB); 1133 } 1134 break; 1135 case UTF16BE: 1136 { 1137 1138 } 1139 break; 1140 default:; 1141 } 1142 } 1143 break; 1144 case UTF16BE: 1145 { 1146 switch (emCodeTypeB) 1147 { 1148 case UTF16LE: 1149 { 1150 return Utf16beFileToUtf16leFile(lpFileA, lpFileB); 1151 } 1152 break; 1153 case UTF8: 1154 { 1155 1156 } 1157 break; 1158 case UTF8_NO_BOM: 1159 { 1160 1161 } 1162 break; 1163 default:; 1164 } 1165 } 1166 break; 1167 default:; 1168 } 1169 1170 return bSUCCESS = TRUE; 1171 } 1172 1173 1174 1175 1176 void test__BYTETOCHAR(); 1177 void test__BYTETOWCHAR(); 1178 1179 int _tmain(int argc, _TCHAR* argv[]) 1180 { 1181 1182 1183 1184 CONST LPTSTR lpInputFileUTF8 = TEXT("input-utf8.txt"); 1185 CONST LPTSTR lpInputFileUTF16le = TEXT("input-utf16le.txt"); 1186 CONST LPTSTR lpInputFileUTF16be = TEXT("input-utf16be.txt"); 1187 CONST LPTSTR lpInputFileUTF8NoBOM = TEXT("input-utf8-no-bom.txt"); 1188 1189 CONST LPTSTR lpOutputFileUTF8NoBOM = TEXT("output-utf8-no-bom.txt"); 1190 CONST LPTSTR lpOutputFileUTF8 = TEXT("output-utf8.txt"); 1191 CONST LPTSTR lpOutputFileUTF16le = TEXT("output-utf16le.txt"); 1192 CONST LPTSTR lpOutputFileUTF16be = TEXT("output-utf16be.txt"); 1193 1194 1195 //CodeFileAToCodeFileB(lpInputFileUTF16le, UTF16LE, lpOutputFileUTF8, UTF8); 1196 //CodeFileAToCodeFileB(lpInputFileUTF16le, UTF16LE, lpOutputFileUTF8NoBOM, UTF8_NO_BOM); 1197 //CodeFileAToCodeFileB(lpInputFileUTF8, UTF8, lpOutputFileUTF8NoBOM, UTF8_NO_BOM); 1198 //CodeFileAToCodeFileB(lpInputFileUTF8, UTF8, lpOutputFileUTF16le, UTF16LE); 1199 //CodeFileAToCodeFileB(lpInputFileUTF8NoBOM, UTF8_NO_BOM, lpOutputFileUTF16le, UTF16LE); 1200 //CodeFileAToCodeFileB(lpInputFileUTF8NoBOM, UTF8_NO_BOM, lpOutputFileUTF8, UTF8); 1201 //CodeFileAToCodeFileB(lpInputFileUTF16le, UTF16LE, lpOutputFileUTF16be, UTF16BE); 1202 //CodeFileAToCodeFileB(lpInputFileUTF16be, UTF16BE, lpOutputFileUTF16le, UTF16LE); 1203 1204 1205 1206 1207 //test__BYTETOCHAR(); 1208 test__BYTETOWCHAR(); 1209 1210 1211 return 0; 1212 } 1213 1214 1215 1216 1217 1218 void test__BYTETOCHAR() 1219 { 1220 //TEXT("output to utf8 with BOM.\r\n输出到utf8带BOM的文件中。\r\nengli sh"); 1221 HANDLE hFileA = NULL; 1222 HANDLE hFileB = NULL; 1223 LPBYTE lpByte = NULL; 1224 INT cbByte = 0; 1225 INT cchByte = 0; 1226 LPSTR lpChar = NULL; 1227 INT cbChar = 0; 1228 INT cchChar = 0; 1229 1230 hFileA = ::CreateFile(TEXT("input.txt"), GENERIC_READ, 0, NULL, OPEN_EXISTING, FILE_ATTRIBUTE_NORMAL, NULL); 1231 if (INVALID_HANDLE_VALUE == hFileA) return; 1232 cchByte = cbByte = SetFilePointer(hFileA, 0, NULL, FILE_END); 1233 if (INVALID_SET_FILE_POINTER == cbByte) return; 1234 lpByte = (BYTE *)malloc(cbByte); 1235 if (NULL == lpByte) return; 1236 ZeroMemory(lpByte, cbByte); 1237 SetFilePointer(hFileA, 0, NULL, FILE_BEGIN); 1238 if (FALSE == ReadFile(hFileA, lpByte, cbByte, NULL, NULL)) return; 1239 1240 cchChar = cbChar = cbByte; 1241 lpChar = (CHAR *)malloc(cbChar); 1242 if (NULL == lpChar) return; 1243 ZeroMemory(lpChar, cbChar); 1244 1245 for (INT i = 0; i != cbByte; ++i) 1246 { 1247 lpChar[i] = (CHAR)lpByte[i]; 1248 } 1249 1250 ////////////////////////////////////////////////////////////////////////// 1251 1252 std::string strText(lpChar); 1253 printf("%s\n", strText.c_str()); 1254 1255 ////////////////////////////////////////////////////////////////////////// 1256 1257 hFileB = ::CreateFile(TEXT("output.txt"), GENERIC_WRITE, 0, NULL, CREATE_ALWAYS, FILE_ATTRIBUTE_NORMAL, NULL); 1258 SetFilePointer(hFileB, 0, NULL, FILE_BEGIN); 1259 if (FALSE == WriteFile(hFileB, lpChar, cbChar, NULL, NULL)) return; 1260 1261 free(lpByte); 1262 free(lpChar); 1263 CloseHandle(hFileA); 1264 CloseHandle(hFileB); 1265 } 1266 void test__BYTETOWCHAR() 1267 { 1268 HANDLE hFileA = NULL; 1269 HANDLE hFileB = NULL; 1270 LPBYTE lpByte = NULL; 1271 INT cbByte = 0; 1272 INT cchByte = 0; 1273 LPWSTR lpWChar = NULL; 1274 INT cbWChar = 0; 1275 INT cchWChar = 0; 1276 1277 hFileA = ::CreateFile(TEXT("input.txt"), GENERIC_READ, 0, NULL, OPEN_EXISTING, FILE_ATTRIBUTE_NORMAL, NULL); 1278 if (INVALID_HANDLE_VALUE == hFileA) return; 1279 cchByte = cbByte = SetFilePointer(hFileA, 0, NULL, FILE_END); 1280 if (INVALID_SET_FILE_POINTER == cbByte) return; 1281 lpByte = (BYTE *)malloc(cbByte); 1282 if (NULL == lpByte) return; 1283 ZeroMemory(lpByte, cbByte); 1284 SetFilePointer(hFileA, 0, NULL, FILE_BEGIN); 1285 if (FALSE == ReadFile(hFileA, lpByte, cbByte, NULL, NULL)) return; 1286 1287 cbWChar = cbByte; 1288 cchWChar = cbWChar / sizeof(WCHAR); 1289 lpWChar = (WCHAR *)malloc(cbWChar); 1290 if (NULL == lpWChar) return; 1291 ZeroMemory(lpWChar, cbWChar); 1292 1293 CopyMemory(lpWChar, lpByte, cbWChar); 1294 1295 ////////////////////////////////////////////////////////////////////////// 1296 1297 std::wstring strText(lpWChar); 1298 wprintf(L"%s\n", strText.c_str()); 1299 1300 ////////////////////////////////////////////////////////////////////////// 1301 1302 hFileB = ::CreateFile(TEXT("output.txt"), GENERIC_WRITE, 0, NULL, CREATE_ALWAYS, FILE_ATTRIBUTE_NORMAL, NULL); 1303 SetFilePointer(hFileB, 0, NULL, FILE_BEGIN); 1304 if (FALSE == WriteFile(hFileB, lpWChar, cbWChar, NULL, NULL)) return; 1305 1306 free(lpByte); 1307 free(lpWChar); 1308 CloseHandle(hFileA); 1309 CloseHandle(hFileB); 1310 } 1311 void test_CHARTOBYTE() 1312 { 1313 1314 } 1315 void test__WCHARTOBYTE() 1316 { 1317 1318 }
*注:由于较长建议大家放到编译器里面阅读。具体编码实例放到这里。
标签:
原文地址:http://www.cnblogs.com/superstargg/p/4248750.html