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 "config.h"
  10 #include "utf8.h"
  11 #include "precompose_utf8.h"
  12
  13 typedef char *iconv_ibp;
  14 static const char *repo_encoding = "UTF-8";
  15 static const char *path_encoding = "UTF-8-MAC";
  16
  17 static size_t has_non_ascii(const char *s, size_t maxlen, size_t *strlen_c)
  18 {
  19         const uint8_t *ptr = (const uint8_t *)s;
  20         size_t strlen_chars = 0;
  21         size_t ret = 0;
  22
  23         if (!ptr || !*ptr)
  24                 return 0;
  25
  26         while (*ptr && maxlen) {
  27                 if (*ptr & 0x80)
  28                         ret++;
  29                 strlen_chars++;
  30                 ptr++;
  31                 maxlen--;
  32         }
  33         if (strlen_c)
  34                 *strlen_c = strlen_chars;
  35
  36         return ret;
  37 }
  38
  39
  40 void probe_utf8_pathname_composition(void)
  41 {
  42         struct strbuf path = STRBUF_INIT;
  43         static const char *auml_nfc = "\xc3\xa4";
  44         static const char *auml_nfd = "\x61\xcc\x88";
  45         int output_fd;
  46         if (precomposed_unicode != -1)
  47                 return; /* We found it defined in the global config, respect it */
  48         git_path_buf(&path, "%s", auml_nfc);
  49         output_fd = open(path.buf, O_CREAT|O_EXCL|O_RDWR, 0600);
  50         if (output_fd >= 0) {
  51                 close(output_fd);
  52                 git_path_buf(&path, "%s", auml_nfd);
  53                 precomposed_unicode = access(path.buf, R_OK) ? 0 : 1;
  54                 git_config_set("core.precomposeunicode",
  55                                precomposed_unicode ? "true" : "false");
  56                 git_path_buf(&path, "%s", auml_nfc);
  57                 if (unlink(path.buf))
  58                         die_errno(_("failed to unlink '%s'"), path.buf);
  59         }
  60         strbuf_release(&path);
  61 }
  62
  63 static inline const char *precompose_string_if_needed(const char *in)
  64 {
  65         size_t inlen;
  66         size_t outlen;
  67         if (has_non_ascii(in, (size_t)-1, &inlen)) {
  68                 iconv_t ic_prec;
  69                 char *out;
  70                 if (precomposed_unicode < 0)
  71                         git_config_get_bool("core.precomposeunicode", &precomposed_unicode);
  72                 if (precomposed_unicode != 1)
  73                         return in;
  74                 ic_prec = iconv_open(repo_encoding, path_encoding);
  75                 if (ic_prec == (iconv_t) -1)
  76                         return in;
  77
  78                 out = reencode_string_iconv(in, inlen, ic_prec, 0, &outlen);
  79                 if (out) {
  80                         if (outlen == inlen && !memcmp(in, out, outlen))
  81                                 free(out); /* no need to return indentical */
  82                         else
  83                                 in = out;
  84                 }
  85                 iconv_close(ic_prec);
  86
  87         }
  88         return in;
  89 }
  90
  91 const char *precompose_argv_prefix(int argc, const char **argv, const char *prefix)
  92 {
  93         int i = 0;
  94
  95         while (i < argc) {
  96                 argv[i] = precompose_string_if_needed(argv[i]);
  97                 i++;
  98         }
  99         if (prefix) {
 100                 prefix = precompose_string_if_needed(prefix);
 101         }
 102         return prefix;
 103 }
 104
 105
 106 PREC_DIR *precompose_utf8_opendir(const char *dirname)
 107 {
 108         PREC_DIR *prec_dir = xmalloc(sizeof(PREC_DIR));
 109         prec_dir->dirent_nfc = xmalloc(sizeof(dirent_prec_psx));
 110         prec_dir->dirent_nfc->max_name_len = sizeof(prec_dir->dirent_nfc->d_name);
 111
 112         prec_dir->dirp = opendir(dirname);
 113         if (!prec_dir->dirp) {
 114                 free(prec_dir->dirent_nfc);
 115                 free(prec_dir);
 116                 return NULL;
 117         } else {
 118                 int ret_errno = errno;
 119                 prec_dir->ic_precompose = iconv_open(repo_encoding, path_encoding);
 120                 /* if iconv_open() fails, die() in readdir() if needed */
 121                 errno = ret_errno;
 122         }
 123
 124         return prec_dir;
 125 }
 126
 127 struct dirent_prec_psx *precompose_utf8_readdir(PREC_DIR *prec_dir)
 128 {
 129         struct dirent *res;
 130         res = readdir(prec_dir->dirp);
 131         if (res) {
 132                 size_t namelenz = strlen(res->d_name) + 1; /* \0 */
 133                 size_t new_maxlen = namelenz;
 134
 135                 int ret_errno = errno;
 136
 137                 if (new_maxlen > prec_dir->dirent_nfc->max_name_len) {
 138                         size_t new_len = sizeof(dirent_prec_psx) + new_maxlen -
 139                                 sizeof(prec_dir->dirent_nfc->d_name);
 140
 141                         prec_dir->dirent_nfc = xrealloc(prec_dir->dirent_nfc, new_len);
 142                         prec_dir->dirent_nfc->max_name_len = new_maxlen;
 143                 }
 144
 145                 prec_dir->dirent_nfc->d_ino  = res->d_ino;
 146                 prec_dir->dirent_nfc->d_type = res->d_type;
 147
 148                 if ((precomposed_unicode == 1) && has_non_ascii(res->d_name, (size_t)-1, NULL)) {
 149                         if (prec_dir->ic_precompose == (iconv_t)-1) {
 150                                 die("iconv_open(%s,%s) failed, but needed:\n"
 151                                                 "    precomposed unicode is not supported.\n"
 152                                                 "    If you want to use decomposed unicode, run\n"
 153                                                 "    \"git config core.precomposeunicode false\"\n",
 154                                                 repo_encoding, path_encoding);
 155                         } else {
 156                                 iconv_ibp       cp = (iconv_ibp)res->d_name;
 157                                 size_t inleft = namelenz;
 158                                 char *outpos = &prec_dir->dirent_nfc->d_name[0];
 159                                 size_t outsz = prec_dir->dirent_nfc->max_name_len;
 160                                 errno = 0;
 161                                 iconv(prec_dir->ic_precompose, &cp, &inleft, &outpos, &outsz);
 162                                 if (errno || inleft) {
 163                                         /*
 164                                          * iconv() failed and errno could be E2BIG, EILSEQ, EINVAL, EBADF
 165                                          * MacOS X avoids illegal byte sequences.
 166                                          * If they occur on a mounted drive (e.g. NFS) it is not worth to
 167                                          * die() for that, but rather let the user see the original name
 168                                         */
 169                                         namelenz = 0; /* trigger strlcpy */
 170                                 }
 171                         }
 172                 } else
 173                         namelenz = 0;
 174
 175                 if (!namelenz)
 176                         strlcpy(prec_dir->dirent_nfc->d_name, res->d_name,
 177                                                         prec_dir->dirent_nfc->max_name_len);
 178
 179                 errno = ret_errno;
 180                 return prec_dir->dirent_nfc;
 181         }
 182         return NULL;
 183 }
 184
 185
 186 int precompose_utf8_closedir(PREC_DIR *prec_dir)
 187 {
 188         int ret_value;
 189         int ret_errno;
 190         ret_value = closedir(prec_dir->dirp);
 191         ret_errno = errno;
 192         if (prec_dir->ic_precompose != (iconv_t)-1)
 193                 iconv_close(prec_dir->ic_precompose);
 194         free(prec_dir->dirent_nfc);
 195         free(prec_dir);
 196         errno = ret_errno;
 197         return ret_value;
 198 }