src/win32/utf-conv.c

   1 /*
   2  * Copyright (C) 2009-2012 the libgit2 contributors
   3  *
   4  * This file is part of libgit2, distributed under the GNU GPL v2 with
   5  * a Linking Exception. For full terms see the included COPYING file.
   6  */
   7
   8 #include "common.h"
   9 #include "utf-conv.h"
  10
  11 #define U16_LEAD(c) (wchar_t)(((c)>>10)+0xd7c0)
  12 #define U16_TRAIL(c) (wchar_t)(((c)&0x3ff)|0xdc00)
  13
  14 #if 0
  15 void git__utf8_to_16(wchar_t *dest, size_t length, const char *src)
  16 {
  17         wchar_t *pDest = dest;
  18         uint32_t ch;
  19         const uint8_t* pSrc = (uint8_t*) src;
  20
  21         assert(dest && src && length);
  22
  23         length--;
  24
  25         while(*pSrc && length > 0) {
  26                 ch = *pSrc++;
  27                 length--;
  28
  29                 if(ch < 0xc0) {
  30                         /*
  31                          * ASCII, or a trail byte in lead position which is treated like
  32                          * a single-byte sequence for better character boundary
  33                          * resynchronization after illegal sequences.
  34                          */
  35                         *pDest++ = (wchar_t)ch;
  36                         continue;
  37                 } else if(ch < 0xe0) { /* U+0080..U+07FF */
  38                         if (pSrc[0]) {
  39                                 /* 0x3080 = (0xc0 << 6) + 0x80 */
  40                                 *pDest++ = (wchar_t)((ch << 6) + *pSrc++ - 0x3080);
  41                                 continue;
  42                         }
  43                 } else if(ch < 0xf0) { /* U+0800..U+FFFF */
  44                         if (pSrc[0] && pSrc[1]) {
  45                                 /* no need for (ch & 0xf) because the upper bits are truncated after <<12 in the cast to (UChar) */
  46                                 /* 0x2080 = (0x80 << 6) + 0x80 */
  47                                 ch = (ch << 12) + (*pSrc++ << 6);
  48                                 *pDest++ = (wchar_t)(ch + *pSrc++ - 0x2080);
  49                                 continue;
  50                         }
  51                 } else /* f0..f4 */ { /* U+10000..U+10FFFF */
  52                         if (length >= 1 && pSrc[0] && pSrc[1] && pSrc[2]) {
  53                                 /* 0x3c82080 = (0xf0 << 18) + (0x80 << 12) + (0x80 << 6) + 0x80 */
  54                                 ch = (ch << 18) + (*pSrc++ << 12);
  55                                 ch += *pSrc++ << 6;
  56                                 ch += *pSrc++ - 0x3c82080;
  57                                 *(pDest++) = U16_LEAD(ch);
  58                                 *(pDest++) = U16_TRAIL(ch);
  59                                 length--; /* two bytes for this character */
  60                                 continue;
  61                         }
  62                 }
  63
  64                 /* truncated character at the end */
  65                 *pDest++ = 0xfffd;
  66                 break;
  67         }
  68
  69         *pDest++ = 0x0;
  70 }
  71 #endif
  72
  73 int git__utf8_to_16(wchar_t *dest, size_t length, const char *src)
  74 {
  75         return MultiByteToWideChar(CP_UTF8, 0, src, -1, dest, (int)length);
  76 }
  77
  78 int git__utf16_to_8(char *out, const wchar_t *input)
  79 {
  80         return WideCharToMultiByte(CP_UTF8, 0, input, -1, out, GIT_WIN_PATH, NULL, NULL);
  81 }