compat/precompose_utf8.c

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