src/win32/utf-conv.c

   1 /*
   2  * Copyright (C) the libgit2 contributors. All rights reserved.
   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 #ifndef WC_ERR_INVALID_CHARS
  12 # define WC_ERR_INVALID_CHARS   0x80
  13 #endif
  14
  15 GIT_INLINE(DWORD) get_wc_flags(void)
  16 {
  17         static char inited = 0;
  18         static DWORD flags;
  19
  20         /* Invalid code point check supported on Vista+ only */
  21         if (!inited) {
  22                 flags = git_has_win32_version(6, 0, 0) ? WC_ERR_INVALID_CHARS : 0;
  23                 inited = 1;
  24         }
  25
  26         return flags;
  27 }
  28
  29 /**
  30  * Converts a UTF-8 string to wide characters.
  31  *
  32  * @param dest The buffer to receive the wide string.
  33  * @param dest_size The size of the buffer, in characters.
  34  * @param src The UTF-8 string to convert.
  35  * @return The length of the wide string, in characters (not counting the NULL terminator), or < 0 for failure
  36  */
  37 int git__utf8_to_16(wchar_t *dest, size_t dest_size, const char *src)
  38 {
  39         /* Length of -1 indicates NULL termination of the input string. Subtract 1 from the result to
  40         * turn 0 into -1 (an error code) and to not count the NULL terminator as part of the string's
  41         * length. MultiByteToWideChar never returns int's minvalue, so underflow is not possible */
  42         return MultiByteToWideChar(CP_UTF8, MB_ERR_INVALID_CHARS, src, -1, dest, (int)dest_size) - 1;
  43 }
  44
  45 /**
  46  * Converts a wide string to UTF-8.
  47  *
  48  * @param dest The buffer to receive the UTF-8 string.
  49  * @param dest_size The size of the buffer, in bytes.
  50  * @param src The wide string to convert.
  51  * @return The length of the UTF-8 string, in bytes (not counting the NULL terminator), or < 0 for failure
  52  */
  53 int git__utf16_to_8(char *dest, size_t dest_size, const wchar_t *src)
  54 {
  55         /* Length of -1 indicates NULL termination of the input string. Subtract 1 from the result to
  56          * turn 0 into -1 (an error code) and to not count the NULL terminator as part of the string's
  57          * length. WideCharToMultiByte never returns int's minvalue, so underflow is not possible */
  58         return WideCharToMultiByte(CP_UTF8, get_wc_flags(), src, -1, dest, (int)dest_size, NULL, NULL) - 1;
  59 }
  60
  61 /**
  62  * Converts a UTF-8 string to wide characters.
  63  * Memory is allocated to hold the converted string.
  64  * The caller is responsible for freeing the string with git__free.
  65  *
  66  * @param dest Receives a pointer to the wide string.
  67  * @param src The UTF-8 string to convert.
  68  * @return The length of the wide string, in characters (not counting the NULL terminator), or < 0 for failure
  69  */
  70 int git__utf8_to_16_alloc(wchar_t **dest, const char *src)
  71 {
  72         int utf16_size;
  73
  74         *dest = NULL;
  75
  76         /* Length of -1 indicates NULL termination of the input string */
  77         utf16_size = MultiByteToWideChar(CP_UTF8, MB_ERR_INVALID_CHARS, src, -1, NULL, 0);
  78
  79         if (!utf16_size)
  80                 return -1;
  81
  82         *dest = git__malloc(utf16_size * sizeof(wchar_t));
  83
  84         if (!*dest)
  85                 return -1;
  86
  87         utf16_size = MultiByteToWideChar(CP_UTF8, MB_ERR_INVALID_CHARS, src, -1, *dest, utf16_size);
  88
  89         if (!utf16_size) {
  90                 git__free(*dest);
  91                 *dest = NULL;
  92         }
  93
  94         /* Subtract 1 from the result to turn 0 into -1 (an error code) and to not count the NULL
  95          * terminator as part of the string's length. MultiByteToWideChar never returns int's minvalue,
  96          * so underflow is not possible */
  97         return utf16_size - 1;
  98 }
  99
 100 /**
 101  * Converts a wide string to UTF-8.
 102  * Memory is allocated to hold the converted string.
 103  * The caller is responsible for freeing the string with git__free.
 104  *
 105  * @param dest Receives a pointer to the UTF-8 string.
 106  * @param src The wide string to convert.
 107  * @return The length of the UTF-8 string, in bytes (not counting the NULL terminator), or < 0 for failure
 108  */
 109 int git__utf16_to_8_alloc(char **dest, const wchar_t *src)
 110 {
 111         int utf8_size;
 112         DWORD dwFlags = get_wc_flags();
 113
 114         *dest = NULL;
 115
 116         /* Length of -1 indicates NULL termination of the input string */
 117         utf8_size = WideCharToMultiByte(CP_UTF8, dwFlags, src, -1, NULL, 0, NULL, NULL);
 118
 119         if (!utf8_size)
 120                 return -1;
 121
 122         *dest = git__malloc(utf8_size);
 123
 124         if (!*dest)
 125                 return -1;
 126
 127         utf8_size = WideCharToMultiByte(CP_UTF8, dwFlags, src, -1, *dest, utf8_size, NULL, NULL);
 128
 129         if (!utf8_size) {
 130                 git__free(*dest);
 131                 *dest = NULL;
 132         }
 133
 134         /* Subtract 1 from the result to turn 0 into -1 (an error code) and to not count the NULL
 135          * terminator as part of the string's length. MultiByteToWideChar never returns int's minvalue,
 136          * so underflow is not possible */
 137         return utf8_size - 1;
 138 }