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 GIT_INLINE(void) git__set_errno(void)
  12 {
  13         if (GetLastError() == ERROR_INSUFFICIENT_BUFFER)
  14                 errno = ENAMETOOLONG;
  15         else
  16                 errno = EINVAL;
  17 }
  18
  19 /**
  20  * Converts a UTF-8 string to wide characters.
  21  *
  22  * @param dest The buffer to receive the wide string.
  23  * @param dest_size The size of the buffer, in characters.
  24  * @param src The UTF-8 string to convert.
  25  * @return The length of the wide string, in characters (not counting the NULL terminator), or < 0 for failure
  26  */
  27 int git__utf8_to_16(wchar_t *dest, size_t dest_size, const char *src)
  28 {
  29         int len;
  30
  31         /* Length of -1 indicates NULL termination of the input string. Subtract 1 from the result to
  32         * turn 0 into -1 (an error code) and to not count the NULL terminator as part of the string's
  33         * length. MultiByteToWideChar never returns int's minvalue, so underflow is not possible */
  34         if ((len = MultiByteToWideChar(CP_UTF8, MB_ERR_INVALID_CHARS, src, -1, dest, (int)dest_size) - 1) < 0)
  35                 git__set_errno();
  36
  37         return len;
  38 }
  39
  40 /**
  41  * Converts a wide string to UTF-8.
  42  *
  43  * @param dest The buffer to receive the UTF-8 string.
  44  * @param dest_size The size of the buffer, in bytes.
  45  * @param src The wide string to convert.
  46  * @return The length of the UTF-8 string, in bytes (not counting the NULL terminator), or < 0 for failure
  47  */
  48 int git__utf16_to_8(char *dest, size_t dest_size, const wchar_t *src)
  49 {
  50         int len;
  51
  52         /* Length of -1 indicates NULL termination of the input string. Subtract 1 from the result to
  53          * turn 0 into -1 (an error code) and to not count the NULL terminator as part of the string's
  54          * length. WideCharToMultiByte never returns int's minvalue, so underflow is not possible */
  55         if ((len = WideCharToMultiByte(CP_UTF8, WC_ERR_INVALID_CHARS, src, -1, dest, (int)dest_size, NULL, NULL) - 1) < 0)
  56                 git__set_errno();
  57
  58         return len;
  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                 git__set_errno();
  81                 return -1;
  82         }
  83
  84         if (!(*dest = git__mallocarray(utf16_size, sizeof(wchar_t)))) {
  85                 errno = ENOMEM;
  86                 return -1;
  87         }
  88
  89         utf16_size = MultiByteToWideChar(CP_UTF8, MB_ERR_INVALID_CHARS, src, -1, *dest, utf16_size);
  90
  91         if (!utf16_size) {
  92                 git__set_errno();
  93
  94                 git__free(*dest);
  95                 *dest = NULL;
  96         }
  97
  98         /* Subtract 1 from the result to turn 0 into -1 (an error code) and to not count the NULL
  99          * terminator as part of the string's length. MultiByteToWideChar never returns int's minvalue,
 100          * so underflow is not possible */
 101         return utf16_size - 1;
 102 }
 103
 104 /**
 105  * Converts a wide string to UTF-8.
 106  * Memory is allocated to hold the converted string.
 107  * The caller is responsible for freeing the string with git__free.
 108  *
 109  * @param dest Receives a pointer to the UTF-8 string.
 110  * @param src The wide string to convert.
 111  * @return The length of the UTF-8 string, in bytes (not counting the NULL terminator), or < 0 for failure
 112  */
 113 int git__utf16_to_8_alloc(char **dest, const wchar_t *src)
 114 {
 115         int utf8_size;
 116
 117         *dest = NULL;
 118
 119         /* Length of -1 indicates NULL termination of the input string */
 120         utf8_size = WideCharToMultiByte(CP_UTF8, WC_ERR_INVALID_CHARS, src, -1, NULL, 0, NULL, NULL);
 121
 122         if (!utf8_size) {
 123                 git__set_errno();
 124                 return -1;
 125         }
 126
 127         *dest = git__malloc(utf8_size);
 128
 129         if (!*dest) {
 130                 errno = ENOMEM;
 131                 return -1;
 132         }
 133
 134         utf8_size = WideCharToMultiByte(CP_UTF8, WC_ERR_INVALID_CHARS, src, -1, *dest, utf8_size, NULL, NULL);
 135
 136         if (!utf8_size) {
 137                 git__set_errno();
 138
 139                 git__free(*dest);
 140                 *dest = NULL;
 141         }
 142
 143         /* Subtract 1 from the result to turn 0 into -1 (an error code) and to not count the NULL
 144          * terminator as part of the string's length. MultiByteToWideChar never returns int's minvalue,
 145          * so underflow is not possible */
 146         return utf8_size - 1;
 147 }