git

1
/*
2
 * Converts filenames from decomposed unicode into precomposed unicode.
3
 * Used on MacOS X.
4
 */
5

6
#define PRECOMPOSE_UNICODE_C
7
#define USE_THE_REPOSITORY_VARIABLE
8

9
#include "git-compat-util.h"
10
#include "config.h"
11
#include "environment.h"
12
#include "gettext.h"
13
#include "path.h"
14
#include "strbuf.h"
15
#include "utf8.h"
16
#include "precompose_utf8.h"
17

18
typedef char *iconv_ibp;
19
static const char *repo_encoding = "UTF-8";
20
static const char *path_encoding = "UTF-8-MAC";
21

22
static size_t has_non_ascii(const char *s, size_t maxlen, size_t *strlen_c)
23
{
24
	const uint8_t *ptr = (const uint8_t *)s;
25
	size_t strlen_chars = 0;
26
	size_t ret = 0;
27

28
	if (!ptr || !*ptr)
29
		return 0;
30

31
	while (*ptr && maxlen) {
32
		if (*ptr & 0x80)
33
			ret++;
34
		strlen_chars++;
35
		ptr++;
36
		maxlen--;
37
	}
38
	if (strlen_c)
39
		*strlen_c = strlen_chars;
40

41
	return ret;
42
}
43

44

45
void probe_utf8_pathname_composition(void)
46
{
47
	struct strbuf path = STRBUF_INIT;
48
	static const char *auml_nfc = "\xc3\xa4";
49
	static const char *auml_nfd = "\x61\xcc\x88";
50
	int output_fd;
51
	if (precomposed_unicode != -1)
52
		return; /* We found it defined in the global config, respect it */
53
	git_path_buf(&path, "%s", auml_nfc);
54
	output_fd = open(path.buf, O_CREAT|O_EXCL|O_RDWR, 0600);
55
	if (output_fd >= 0) {
56
		close(output_fd);
57
		git_path_buf(&path, "%s", auml_nfd);
58
		precomposed_unicode = access(path.buf, R_OK) ? 0 : 1;
59
		git_config_set("core.precomposeunicode",
60
			       precomposed_unicode ? "true" : "false");
61
		git_path_buf(&path, "%s", auml_nfc);
62
		if (unlink(path.buf))
63
			die_errno(_("failed to unlink '%s'"), path.buf);
64
	}
65
	strbuf_release(&path);
66
}
67

68
const char *precompose_string_if_needed(const char *in)
69
{
70
	size_t inlen;
71
	size_t outlen;
72
	if (!in)
73
		return NULL;
74
	if (has_non_ascii(in, (size_t)-1, &inlen)) {
75
		iconv_t ic_prec;
76
		char *out;
77
		if (precomposed_unicode < 0)
78
			git_config_get_bool("core.precomposeunicode", &precomposed_unicode);
79
		if (precomposed_unicode != 1)
80
			return in;
81
		ic_prec = iconv_open(repo_encoding, path_encoding);
82
		if (ic_prec == (iconv_t) -1)
83
			return in;
84

85
		out = reencode_string_iconv(in, inlen, ic_prec, 0, &outlen);
86
		if (out) {
87
			if (outlen == inlen && !memcmp(in, out, outlen))
88
				free(out); /* no need to return indentical */
89
			else
90
				in = out;
91
		}
92
		iconv_close(ic_prec);
93

94
	}
95
	return in;
96
}
97

98
const char *precompose_argv_prefix(int argc, const char **argv, const char *prefix)
99
{
100
	int i = 0;
101

102
	while (i < argc) {
103
		argv[i] = precompose_string_if_needed(argv[i]);
104
		i++;
105
	}
106
	return precompose_string_if_needed(prefix);
107
}
108

109

110
PREC_DIR *precompose_utf8_opendir(const char *dirname)
111
{
112
	PREC_DIR *prec_dir = xmalloc(sizeof(PREC_DIR));
113
	prec_dir->dirent_nfc = xmalloc(sizeof(dirent_prec_psx));
114
	prec_dir->dirent_nfc->max_name_len = sizeof(prec_dir->dirent_nfc->d_name);
115

116
	prec_dir->dirp = opendir(dirname);
117
	if (!prec_dir->dirp) {
118
		free(prec_dir->dirent_nfc);
119
		free(prec_dir);
120
		return NULL;
121
	} else {
122
		int ret_errno = errno;
123
		prec_dir->ic_precompose = iconv_open(repo_encoding, path_encoding);
124
		/* if iconv_open() fails, die() in readdir() if needed */
125
		errno = ret_errno;
126
	}
127

128
	return prec_dir;
129
}
130

131
struct dirent_prec_psx *precompose_utf8_readdir(PREC_DIR *prec_dir)
132
{
133
	struct dirent *res;
134
	res = readdir(prec_dir->dirp);
135
	if (res) {
136
		size_t namelenz = strlen(res->d_name) + 1; /* \0 */
137
		size_t new_maxlen = namelenz;
138

139
		int ret_errno = errno;
140

141
		if (new_maxlen > prec_dir->dirent_nfc->max_name_len) {
142
			size_t new_len = sizeof(dirent_prec_psx) + new_maxlen -
143
				sizeof(prec_dir->dirent_nfc->d_name);
144

145
			prec_dir->dirent_nfc = xrealloc(prec_dir->dirent_nfc, new_len);
146
			prec_dir->dirent_nfc->max_name_len = new_maxlen;
147
		}
148

149
		prec_dir->dirent_nfc->d_ino  = res->d_ino;
150
		prec_dir->dirent_nfc->d_type = res->d_type;
151

152
		if ((precomposed_unicode == 1) && has_non_ascii(res->d_name, (size_t)-1, NULL)) {
153
			if (prec_dir->ic_precompose == (iconv_t)-1) {
154
				die("iconv_open(%s,%s) failed, but needed:\n"
155
						"    precomposed unicode is not supported.\n"
156
						"    If you want to use decomposed unicode, run\n"
157
						"    \"git config core.precomposeunicode false\"\n",
158
						repo_encoding, path_encoding);
159
			} else {
160
				iconv_ibp	cp = (iconv_ibp)res->d_name;
161
				size_t inleft = namelenz;
162
				char *outpos = &prec_dir->dirent_nfc->d_name[0];
163
				size_t outsz = prec_dir->dirent_nfc->max_name_len;
164
				errno = 0;
165
				iconv(prec_dir->ic_precompose, &cp, &inleft, &outpos, &outsz);
166
				if (errno || inleft) {
167
					/*
168
					 * iconv() failed and errno could be E2BIG, EILSEQ, EINVAL, EBADF
169
					 * MacOS X avoids illegal byte sequences.
170
					 * If they occur on a mounted drive (e.g. NFS) it is not worth to
171
					 * die() for that, but rather let the user see the original name
172
					*/
173
					namelenz = 0; /* trigger strlcpy */
174
				}
175
			}
176
		} else
177
			namelenz = 0;
178

179
		if (!namelenz)
180
			strlcpy(prec_dir->dirent_nfc->d_name, res->d_name,
181
							prec_dir->dirent_nfc->max_name_len);
182

183
		errno = ret_errno;
184
		return prec_dir->dirent_nfc;
185
	}
186
	return NULL;
187
}
188

189

190
int precompose_utf8_closedir(PREC_DIR *prec_dir)
191
{
192
	int ret_value;
193
	int ret_errno;
194
	ret_value = closedir(prec_dir->dirp);
195
	ret_errno = errno;
196
	if (prec_dir->ic_precompose != (iconv_t)-1)
197
		iconv_close(prec_dir->ic_precompose);
198
	free(prec_dir->dirent_nfc);
199
	free(prec_dir);
200
	errno = ret_errno;
201
	return ret_value;
202
}
203