compat/precompose_utf8.c

   1 /*
   2  * Converts filenames from decomposed unicode into precomposed unicode.
   3  * Used on MacOS X.
   4 */
   5
   6
   7 #define PRECOMPOSE_UNICODE_C
   8
   9 #include "cache.h"
  10 #include "utf8.h"
  11 #include "precompose_utf8.h"
  12
  13 typedef char *iconv_ibp;
  14 const static char *repo_encoding = "UTF-8";
  15 const static char *path_encoding = "UTF-8-MAC";
  16
  17
  18 static size_t has_utf8(const char *s, size_t maxlen, size_t *strlen_c)
  19 {
  20         const uint8_t *utf8p = (const uint8_t*) s;
  21         size_t strlen_chars = 0;
  22         size_t ret = 0;
  23
  24         if ((!utf8p) || (!*utf8p)) {
  25                 return 0;
  26         }
  27
  28         while((*utf8p) && maxlen) {
  29                 if (*utf8p & 0x80)
  30                         ret++;
  31                 strlen_chars++;
  32                 utf8p++;
  33                 maxlen--;
  34         }
  35         if (strlen_c)
  36                 *strlen_c = strlen_chars;
  37
  38         return ret;
  39 }
  40
  41
  42 void probe_utf8_pathname_composition(char *path, int len)
  43 {
  44         const static char *auml_nfc = "\xc3\xa4";
  45         const static char *auml_nfd = "\x61\xcc\x88";
  46         int output_fd;
  47         if (precomposed_unicode != -1)
  48                 return; /* We found it defined in the global config, respect it */
  49         path[len] = 0;
  50         strcpy(path + len, auml_nfc);
  51         output_fd = open(path, O_CREAT|O_EXCL|O_RDWR, 0600);
  52         if (output_fd >=0) {
  53                 close(output_fd);
  54                 path[len] = 0;
  55                 strcpy(path + len, auml_nfd);
  56                 /* Indicate to the user, that we can configure it to true */
  57                 if (0 == access(path, R_OK))
  58                         git_config_set("core.precomposeunicode", "false");
  59                         /* To be backward compatible, set precomposed_unicode to 0 */
  60                 precomposed_unicode = 0;
  61                 path[len] = 0;
  62                 strcpy(path + len, auml_nfc);
  63                 unlink(path);
  64         }
  65 }
  66
  67
  68 void precompose_argv(int argc, const char **argv)
  69 {
  70         int i = 0;
  71         const char *oldarg;
  72         char *newarg;
  73         iconv_t ic_precompose;
  74
  75         if (precomposed_unicode != 1)
  76                 return;
  77
  78         ic_precompose = iconv_open(repo_encoding, path_encoding);
  79         if (ic_precompose == (iconv_t) -1)
  80                 return;
  81
  82         while (i < argc) {
  83                 size_t namelen;
  84                 oldarg = argv[i];
  85                 if (has_utf8(oldarg, (size_t)-1, &namelen)) {
  86                         newarg = reencode_string_iconv(oldarg, namelen, ic_precompose);
  87                         if (newarg)
  88                                 argv[i] = newarg;
  89                 }
  90                 i++;
  91         }
  92         iconv_close(ic_precompose);
  93 }
  94
  95
  96 PREC_DIR *precompose_utf8_opendir(const char *dirname)
  97 {
  98         PREC_DIR *prec_dir = xmalloc(sizeof(PREC_DIR));
  99         prec_dir->dirent_nfc = xmalloc(sizeof(dirent_prec_psx));
 100         prec_dir->dirent_nfc->max_name_len = sizeof(prec_dir->dirent_nfc->d_name);
 101
 102         prec_dir->dirp = opendir(dirname);
 103         if (!prec_dir->dirp) {
 104                 free(prec_dir->dirent_nfc);
 105                 free(prec_dir);
 106                 return NULL;
 107         } else {
 108                 int ret_errno = errno;
 109                 prec_dir->ic_precompose = iconv_open(repo_encoding, path_encoding);
 110                 /* if iconv_open() fails, die() in readdir() if needed */
 111                 errno = ret_errno;
 112         }
 113
 114         return prec_dir;
 115 }
 116
 117 struct dirent_prec_psx *precompose_utf8_readdir(PREC_DIR *prec_dir)
 118 {
 119         struct dirent *res;
 120         res = readdir(prec_dir->dirp);
 121         if (res) {
 122                 size_t namelenz = strlen(res->d_name) + 1; /* \0 */
 123                 size_t new_maxlen = namelenz;
 124
 125                 int ret_errno = errno;
 126
 127                 if (new_maxlen > prec_dir->dirent_nfc->max_name_len) {
 128                         size_t new_len = sizeof(dirent_prec_psx) + new_maxlen -
 129                                 sizeof(prec_dir->dirent_nfc->d_name);
 130
 131                         prec_dir->dirent_nfc = xrealloc(prec_dir->dirent_nfc, new_len);
 132                         prec_dir->dirent_nfc->max_name_len = new_maxlen;
 133                 }
 134
 135                 prec_dir->dirent_nfc->d_ino  = res->d_ino;
 136                 prec_dir->dirent_nfc->d_type = res->d_type;
 137
 138                 if ((precomposed_unicode == 1) && has_utf8(res->d_name, (size_t)-1, NULL)) {
 139                         if (prec_dir->ic_precompose == (iconv_t)-1) {
 140                                 die("iconv_open(%s,%s) failed, but needed:\n"
 141                                                 "    precomposed unicode is not supported.\n"
 142                                                 "    If you wnat to use decomposed unicode, run\n"
 143                                                 "    \"git config core.precomposeunicode false\"\n",
 144                                                 repo_encoding, path_encoding);
 145                         } else {
 146                                 iconv_ibp       cp = (iconv_ibp)res->d_name;
 147                                 size_t inleft = namelenz;
 148                                 char *outpos = &prec_dir->dirent_nfc->d_name[0];
 149                                 size_t outsz = prec_dir->dirent_nfc->max_name_len;
 150                                 size_t cnt;
 151                                 errno = 0;
 152                                 cnt = iconv(prec_dir->ic_precompose, &cp, &inleft, &outpos, &outsz);
 153                                 if (errno || inleft) {
 154                                         /*
 155                                          * iconv() failed and errno could be E2BIG, EILSEQ, EINVAL, EBADF
 156                                          * MacOS X avoids illegal byte sequemces.
 157                                          * If they occur on a mounted drive (e.g. NFS) it is not worth to
 158                                          * die() for that, but rather let the user see the original name
 159                                         */
 160                                         namelenz = 0; /* trigger strlcpy */
 161                                 }
 162                         }
 163                 }
 164                 else
 165                         namelenz = 0;
 166
 167                 if (!namelenz)
 168                         strlcpy(prec_dir->dirent_nfc->d_name, res->d_name,
 169                                                         prec_dir->dirent_nfc->max_name_len);
 170
 171                 errno = ret_errno;
 172                 return prec_dir->dirent_nfc;
 173         }
 174         return NULL;
 175 }
 176
 177
 178 int precompose_utf8_closedir(PREC_DIR *prec_dir)
 179 {
 180         int ret_value;
 181         int ret_errno;
 182         ret_value = closedir(prec_dir->dirp);
 183         ret_errno = errno;
 184         if (prec_dir->ic_precompose != (iconv_t)-1)
 185                 iconv_close(prec_dir->ic_precompose);
 186         free(prec_dir->dirent_nfc);
 187         free(prec_dir);
 188         errno = ret_errno;
 189         return ret_value;
 190 }