src/util/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 "utf-conv.h"
   9
  10 GIT_INLINE(void) git__set_errno(void)
  11 {
  12         if (GetLastError() == ERROR_INSUFFICIENT_BUFFER)
  13                 errno = ENAMETOOLONG;
  14         else
  15                 errno = EINVAL;
  16 }
  17
  18 /**
  19  * Converts a UTF-8 string to wide characters.
  20  *
  21  * @param dest The buffer to receive the wide string.
  22  * @param dest_size The size of the buffer, in characters.
  23  * @param src The UTF-8 string to convert.
  24  * @return The length of the wide string, in characters (not counting the NULL terminator), or < 0 for failure
  25  */
  26 int git__utf8_to_16(wchar_t *dest, size_t dest_size, const char *src)
  27 {
  28         int len;
  29
  30         /* Length of -1 indicates NULL termination of the input string. Subtract 1 from the result to
  31         * turn 0 into -1 (an error code) and to not count the NULL terminator as part of the string's
  32         * length. MultiByteToWideChar never returns int's minvalue, so underflow is not possible */
  33         if ((len = MultiByteToWideChar(CP_UTF8, MB_ERR_INVALID_CHARS, src, -1, dest, (int)dest_size) - 1) < 0)
  34                 git__set_errno();
  35
  36         return len;
  37 }
  38
  39 /**
  40  * Converts a wide string to UTF-8.
  41  *
  42  * @param dest The buffer to receive the UTF-8 string.
  43  * @param dest_size The size of the buffer, in bytes.
  44  * @param src The wide string to convert.
  45  * @return The length of the UTF-8 string, in bytes (not counting the NULL terminator), or < 0 for failure
  46  */
  47 int git__utf16_to_8(char *dest, size_t dest_size, const wchar_t *src)
  48 {
  49         int len;
  50
  51         /* Length of -1 indicates NULL termination of the input string. Subtract 1 from the result to
  52          * turn 0 into -1 (an error code) and to not count the NULL terminator as part of the string's
  53          * length. WideCharToMultiByte never returns int's minvalue, so underflow is not possible */
  54         if ((len = WideCharToMultiByte(CP_UTF8, WC_ERR_INVALID_CHARS, src, -1, dest, (int)dest_size, NULL, NULL) - 1) < 0)
  55                 git__set_errno();
  56
  57         return len;
  58 }
  59
  60 /**
  61  * Converts a UTF-8 string to wide characters.
  62  * Memory is allocated to hold the converted string.
  63  * The caller is responsible for freeing the string with git__free.
  64  *
  65  * @param dest Receives a pointer to the wide string.
  66  * @param src The UTF-8 string to convert.
  67  * @return The length of the wide string, in characters (not counting the NULL terminator), or < 0 for failure
  68  */
  69 int git__utf8_to_16_alloc(wchar_t **dest, const char *src)
  70 {
  71         int utf16_size;
  72
  73         *dest = NULL;
  74
  75         /* Length of -1 indicates NULL termination of the input string */
  76         utf16_size = MultiByteToWideChar(CP_UTF8, MB_ERR_INVALID_CHARS, src, -1, NULL, 0);
  77
  78         if (!utf16_size) {
  79                 git__set_errno();
  80                 return -1;
  81         }
  82
  83         if (!(*dest = git__mallocarray(utf16_size, sizeof(wchar_t)))) {
  84                 errno = ENOMEM;
  85                 return -1;
  86         }
  87
  88         utf16_size = MultiByteToWideChar(CP_UTF8, MB_ERR_INVALID_CHARS, src, -1, *dest, utf16_size);
  89
  90         if (!utf16_size) {
  91                 git__set_errno();
  92
  93                 git__free(*dest);
  94                 *dest = NULL;
  95         }
  96
  97         /* Subtract 1 from the result to turn 0 into -1 (an error code) and to not count the NULL
  98          * terminator as part of the string's length. MultiByteToWideChar never returns int's minvalue,
  99          * so underflow is not possible */
 100         return utf16_size - 1;
 101 }
 102
 103 /**
 104  * Converts a wide string to UTF-8.
 105  * Memory is allocated to hold the converted string.
 106  * The caller is responsible for freeing the string with git__free.
 107  *
 108  * @param dest Receives a pointer to the UTF-8 string.
 109  * @param src The wide string to convert.
 110  * @return The length of the UTF-8 string, in bytes (not counting the NULL terminator), or < 0 for failure
 111  */
 112 int git__utf16_to_8_alloc(char **dest, const wchar_t *src)
 113 {
 114         int utf8_size;
 115
 116         *dest = NULL;
 117
 118         /* Length of -1 indicates NULL termination of the input string */
 119         utf8_size = WideCharToMultiByte(CP_UTF8, WC_ERR_INVALID_CHARS, src, -1, NULL, 0, NULL, NULL);
 120
 121         if (!utf8_size) {
 122                 git__set_errno();
 123                 return -1;
 124         }
 125
 126         *dest = git__malloc(utf8_size);
 127
 128         if (!*dest) {
 129                 errno = ENOMEM;
 130                 return -1;
 131         }
 132
 133         utf8_size = WideCharToMultiByte(CP_UTF8, WC_ERR_INVALID_CHARS, src, -1, *dest, utf8_size, NULL, NULL);
 134
 135         if (!utf8_size) {
 136                 git__set_errno();
 137
 138                 git__free(*dest);
 139                 *dest = NULL;
 140         }
 141
 142         /* Subtract 1 from the result to turn 0 into -1 (an error code) and to not count the NULL
 143          * terminator as part of the string's length. MultiByteToWideChar never returns int's minvalue,
 144          * so underflow is not possible */
 145         return utf8_size - 1;
 146 }