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 
has_non_ascii(const char * s,size_t maxlen,size_t * strlen_c)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 
probe_utf8_pathname_composition(void)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 
precompose_string_if_needed(const char * in)63 const char *precompose_string_if_needed(const char *in)
64 {
65 	size_t inlen;
66 	size_t outlen;
67 	if (!in)
68 		return NULL;
69 	if (has_non_ascii(in, (size_t)-1, &inlen)) {
70 		iconv_t ic_prec;
71 		char *out;
72 		if (precomposed_unicode < 0)
73 			git_config_get_bool("core.precomposeunicode", &precomposed_unicode);
74 		if (precomposed_unicode != 1)
75 			return in;
76 		ic_prec = iconv_open(repo_encoding, path_encoding);
77 		if (ic_prec == (iconv_t) -1)
78 			return in;
79 
80 		out = reencode_string_iconv(in, inlen, ic_prec, 0, &outlen);
81 		if (out) {
82 			if (outlen == inlen && !memcmp(in, out, outlen))
83 				free(out); /* no need to return indentical */
84 			else
85 				in = out;
86 		}
87 		iconv_close(ic_prec);
88 
89 	}
90 	return in;
91 }
92 
precompose_argv_prefix(int argc,const char ** argv,const char * prefix)93 const char *precompose_argv_prefix(int argc, const char **argv, const char *prefix)
94 {
95 	int i = 0;
96 
97 	while (i < argc) {
98 		argv[i] = precompose_string_if_needed(argv[i]);
99 		i++;
100 	}
101 	return precompose_string_if_needed(prefix);
102 }
103 
104 
precompose_utf8_opendir(const char * dirname)105 PREC_DIR *precompose_utf8_opendir(const char *dirname)
106 {
107 	PREC_DIR *prec_dir = xmalloc(sizeof(PREC_DIR));
108 	prec_dir->dirent_nfc = xmalloc(sizeof(dirent_prec_psx));
109 	prec_dir->dirent_nfc->max_name_len = sizeof(prec_dir->dirent_nfc->d_name);
110 
111 	prec_dir->dirp = opendir(dirname);
112 	if (!prec_dir->dirp) {
113 		free(prec_dir->dirent_nfc);
114 		free(prec_dir);
115 		return NULL;
116 	} else {
117 		int ret_errno = errno;
118 		prec_dir->ic_precompose = iconv_open(repo_encoding, path_encoding);
119 		/* if iconv_open() fails, die() in readdir() if needed */
120 		errno = ret_errno;
121 	}
122 
123 	return prec_dir;
124 }
125 
precompose_utf8_readdir(PREC_DIR * prec_dir)126 struct dirent_prec_psx *precompose_utf8_readdir(PREC_DIR *prec_dir)
127 {
128 	struct dirent *res;
129 	res = readdir(prec_dir->dirp);
130 	if (res) {
131 		size_t namelenz = strlen(res->d_name) + 1; /* \0 */
132 		size_t new_maxlen = namelenz;
133 
134 		int ret_errno = errno;
135 
136 		if (new_maxlen > prec_dir->dirent_nfc->max_name_len) {
137 			size_t new_len = sizeof(dirent_prec_psx) + new_maxlen -
138 				sizeof(prec_dir->dirent_nfc->d_name);
139 
140 			prec_dir->dirent_nfc = xrealloc(prec_dir->dirent_nfc, new_len);
141 			prec_dir->dirent_nfc->max_name_len = new_maxlen;
142 		}
143 
144 		prec_dir->dirent_nfc->d_ino  = res->d_ino;
145 		prec_dir->dirent_nfc->d_type = res->d_type;
146 
147 		if ((precomposed_unicode == 1) && has_non_ascii(res->d_name, (size_t)-1, NULL)) {
148 			if (prec_dir->ic_precompose == (iconv_t)-1) {
149 				die("iconv_open(%s,%s) failed, but needed:\n"
150 						"    precomposed unicode is not supported.\n"
151 						"    If you want to use decomposed unicode, run\n"
152 						"    \"git config core.precomposeunicode false\"\n",
153 						repo_encoding, path_encoding);
154 			} else {
155 				iconv_ibp	cp = (iconv_ibp)res->d_name;
156 				size_t inleft = namelenz;
157 				char *outpos = &prec_dir->dirent_nfc->d_name[0];
158 				size_t outsz = prec_dir->dirent_nfc->max_name_len;
159 				errno = 0;
160 				iconv(prec_dir->ic_precompose, &cp, &inleft, &outpos, &outsz);
161 				if (errno || inleft) {
162 					/*
163 					 * iconv() failed and errno could be E2BIG, EILSEQ, EINVAL, EBADF
164 					 * MacOS X avoids illegal byte sequences.
165 					 * If they occur on a mounted drive (e.g. NFS) it is not worth to
166 					 * die() for that, but rather let the user see the original name
167 					*/
168 					namelenz = 0; /* trigger strlcpy */
169 				}
170 			}
171 		} else
172 			namelenz = 0;
173 
174 		if (!namelenz)
175 			strlcpy(prec_dir->dirent_nfc->d_name, res->d_name,
176 							prec_dir->dirent_nfc->max_name_len);
177 
178 		errno = ret_errno;
179 		return prec_dir->dirent_nfc;
180 	}
181 	return NULL;
182 }
183 
184 
precompose_utf8_closedir(PREC_DIR * prec_dir)185 int precompose_utf8_closedir(PREC_DIR *prec_dir)
186 {
187 	int ret_value;
188 	int ret_errno;
189 	ret_value = closedir(prec_dir->dirp);
190 	ret_errno = errno;
191 	if (prec_dir->ic_precompose != (iconv_t)-1)
192 		iconv_close(prec_dir->ic_precompose);
193 	free(prec_dir->dirent_nfc);
194 	free(prec_dir);
195 	errno = ret_errno;
196 	return ret_value;
197 }
198