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