blob: 44b815e57f9439116199f91b67489a98272272bc [file] [log] [blame]
Linus Torvalds1da177e2005-04-16 15:20:36 -07001/*
2 * unicode.c
3 *
4 * PURPOSE
5 * Routines for converting between UTF-8 and OSTA Compressed Unicode.
6 * Also handles filename mangling
7 *
8 * DESCRIPTION
9 * OSTA Compressed Unicode is explained in the OSTA UDF specification.
10 * http://www.osta.org/
11 * UTF-8 is explained in the IETF RFC XXXX.
12 * ftp://ftp.internic.net/rfc/rfcxxxx.txt
13 *
Linus Torvalds1da177e2005-04-16 15:20:36 -070014 * COPYRIGHT
15 * This file is distributed under the terms of the GNU General Public
16 * License (GPL). Copies of the GPL can be obtained from:
17 * ftp://prep.ai.mit.edu/pub/gnu/GPL
18 * Each contributing author retains all rights to their own work.
19 */
20
21#include "udfdecl.h"
22
23#include <linux/kernel.h>
24#include <linux/string.h> /* for memset */
25#include <linux/nls.h>
Bob Copelandf845fce2008-04-17 09:47:48 +020026#include <linux/crc-itu-t.h>
Tejun Heo5a0e3ad2010-03-24 17:04:11 +090027#include <linux/slab.h>
Linus Torvalds1da177e2005-04-16 15:20:36 -070028
29#include "udf_sb.h"
30
31static int udf_translate_to_linux(uint8_t *, uint8_t *, int, uint8_t *, int);
32
Cyrill Gorcunov28de7942007-07-21 04:37:18 -070033static int udf_char_to_ustr(struct ustr *dest, const uint8_t *src, int strlen)
Linus Torvalds1da177e2005-04-16 15:20:36 -070034{
Cyrill Gorcunovcb00ea32007-07-19 01:47:43 -070035 if ((!dest) || (!src) || (!strlen) || (strlen > UDF_NAME_LEN - 2))
Linus Torvalds1da177e2005-04-16 15:20:36 -070036 return 0;
Cyrill Gorcunov28de7942007-07-21 04:37:18 -070037
Linus Torvalds1da177e2005-04-16 15:20:36 -070038 memset(dest, 0, sizeof(struct ustr));
39 memcpy(dest->u_name, src, strlen);
40 dest->u_cmpID = 0x08;
41 dest->u_len = strlen;
Cyrill Gorcunov28de7942007-07-21 04:37:18 -070042
Linus Torvalds1da177e2005-04-16 15:20:36 -070043 return strlen;
44}
45
46/*
47 * udf_build_ustr
48 */
Cyrill Gorcunov28de7942007-07-21 04:37:18 -070049int udf_build_ustr(struct ustr *dest, dstring *ptr, int size)
Linus Torvalds1da177e2005-04-16 15:20:36 -070050{
51 int usesize;
52
Marcin Slusarz6305a0a2008-02-04 22:27:39 +010053 if (!dest || !ptr || !size)
Linus Torvalds1da177e2005-04-16 15:20:36 -070054 return -1;
Marcin Slusarz6305a0a2008-02-04 22:27:39 +010055 BUG_ON(size < 2);
Linus Torvalds1da177e2005-04-16 15:20:36 -070056
Marcin Slusarz6305a0a2008-02-04 22:27:39 +010057 usesize = min_t(size_t, ptr[size - 1], sizeof(dest->u_name));
58 usesize = min(usesize, size - 2);
Cyrill Gorcunovcb00ea32007-07-19 01:47:43 -070059 dest->u_cmpID = ptr[0];
Marcin Slusarz6305a0a2008-02-04 22:27:39 +010060 dest->u_len = usesize;
61 memcpy(dest->u_name, ptr + 1, usesize);
62 memset(dest->u_name + usesize, 0, sizeof(dest->u_name) - usesize);
Cyrill Gorcunov28de7942007-07-21 04:37:18 -070063
Linus Torvalds1da177e2005-04-16 15:20:36 -070064 return 0;
65}
66
67/*
68 * udf_build_ustr_exact
69 */
Cyrill Gorcunov28de7942007-07-21 04:37:18 -070070static int udf_build_ustr_exact(struct ustr *dest, dstring *ptr, int exactsize)
Linus Torvalds1da177e2005-04-16 15:20:36 -070071{
Cyrill Gorcunovcb00ea32007-07-19 01:47:43 -070072 if ((!dest) || (!ptr) || (!exactsize))
Linus Torvalds1da177e2005-04-16 15:20:36 -070073 return -1;
74
75 memset(dest, 0, sizeof(struct ustr));
Cyrill Gorcunovcb00ea32007-07-19 01:47:43 -070076 dest->u_cmpID = ptr[0];
77 dest->u_len = exactsize - 1;
78 memcpy(dest->u_name, ptr + 1, exactsize - 1);
Cyrill Gorcunov28de7942007-07-21 04:37:18 -070079
Linus Torvalds1da177e2005-04-16 15:20:36 -070080 return 0;
81}
82
83/*
84 * udf_ocu_to_utf8
85 *
86 * PURPOSE
87 * Convert OSTA Compressed Unicode to the UTF-8 equivalent.
88 *
Linus Torvalds1da177e2005-04-16 15:20:36 -070089 * PRE-CONDITIONS
90 * utf Pointer to UTF-8 output buffer.
91 * ocu Pointer to OSTA Compressed Unicode input buffer
92 * of size UDF_NAME_LEN bytes.
93 * both of type "struct ustr *"
94 *
95 * POST-CONDITIONS
96 * <return> Zero on success.
97 *
98 * HISTORY
99 * November 12, 1997 - Andrew E. Mileski
100 * Written, tested, and released.
101 */
marcin.slusarz@gmail.com79cfe0f2008-01-30 22:03:51 +0100102int udf_CS0toUTF8(struct ustr *utf_o, const struct ustr *ocu_i)
Linus Torvalds1da177e2005-04-16 15:20:36 -0700103{
marcin.slusarz@gmail.com79cfe0f2008-01-30 22:03:51 +0100104 const uint8_t *ocu;
Linus Torvalds1da177e2005-04-16 15:20:36 -0700105 uint8_t cmp_id, ocu_len;
106 int i;
107
Linus Torvalds1da177e2005-04-16 15:20:36 -0700108 ocu_len = ocu_i->u_len;
Cyrill Gorcunovcb00ea32007-07-19 01:47:43 -0700109 if (ocu_len == 0) {
Linus Torvalds1da177e2005-04-16 15:20:36 -0700110 memset(utf_o, 0, sizeof(struct ustr));
Linus Torvalds1da177e2005-04-16 15:20:36 -0700111 return 0;
112 }
113
marcin.slusarz@gmail.com79cfe0f2008-01-30 22:03:51 +0100114 cmp_id = ocu_i->u_cmpID;
115 if (cmp_id != 8 && cmp_id != 16) {
116 memset(utf_o, 0, sizeof(struct ustr));
Joe Perches78ace702011-10-10 01:08:05 -0700117 pr_err("unknown compression code (%d) stri=%s\n",
Cyrill Gorcunovcb00ea32007-07-19 01:47:43 -0700118 cmp_id, ocu_i->u_name);
Linus Torvalds1da177e2005-04-16 15:20:36 -0700119 return 0;
120 }
121
marcin.slusarz@gmail.com79cfe0f2008-01-30 22:03:51 +0100122 ocu = ocu_i->u_name;
123 utf_o->u_len = 0;
Cyrill Gorcunovcb00ea32007-07-19 01:47:43 -0700124 for (i = 0; (i < ocu_len) && (utf_o->u_len <= (UDF_NAME_LEN - 3));) {
Linus Torvalds1da177e2005-04-16 15:20:36 -0700125
126 /* Expand OSTA compressed Unicode to Unicode */
marcin.slusarz@gmail.com79cfe0f2008-01-30 22:03:51 +0100127 uint32_t c = ocu[i++];
Linus Torvalds1da177e2005-04-16 15:20:36 -0700128 if (cmp_id == 16)
129 c = (c << 8) | ocu[i++];
130
131 /* Compress Unicode to UTF-8 */
marcin.slusarz@gmail.com79cfe0f2008-01-30 22:03:51 +0100132 if (c < 0x80U)
Cyrill Gorcunov28de7942007-07-21 04:37:18 -0700133 utf_o->u_name[utf_o->u_len++] = (uint8_t)c;
marcin.slusarz@gmail.com79cfe0f2008-01-30 22:03:51 +0100134 else if (c < 0x800U) {
Marcin Slusarz4b111112008-02-08 04:20:36 -0800135 utf_o->u_name[utf_o->u_len++] =
136 (uint8_t)(0xc0 | (c >> 6));
137 utf_o->u_name[utf_o->u_len++] =
138 (uint8_t)(0x80 | (c & 0x3f));
Cyrill Gorcunovcb00ea32007-07-19 01:47:43 -0700139 } else {
Marcin Slusarz4b111112008-02-08 04:20:36 -0800140 utf_o->u_name[utf_o->u_len++] =
141 (uint8_t)(0xe0 | (c >> 12));
142 utf_o->u_name[utf_o->u_len++] =
143 (uint8_t)(0x80 |
144 ((c >> 6) & 0x3f));
145 utf_o->u_name[utf_o->u_len++] =
146 (uint8_t)(0x80 | (c & 0x3f));
Linus Torvalds1da177e2005-04-16 15:20:36 -0700147 }
148 }
Cyrill Gorcunovcb00ea32007-07-19 01:47:43 -0700149 utf_o->u_cmpID = 8;
Linus Torvalds1da177e2005-04-16 15:20:36 -0700150
151 return utf_o->u_len;
152}
153
154/*
155 *
156 * udf_utf8_to_ocu
157 *
158 * PURPOSE
159 * Convert UTF-8 to the OSTA Compressed Unicode equivalent.
160 *
161 * DESCRIPTION
162 * This routine is only called by udf_lookup().
163 *
164 * PRE-CONDITIONS
165 * ocu Pointer to OSTA Compressed Unicode output
166 * buffer of size UDF_NAME_LEN bytes.
167 * utf Pointer to UTF-8 input buffer.
168 * utf_len Length of UTF-8 input buffer in bytes.
169 *
170 * POST-CONDITIONS
171 * <return> Zero on success.
172 *
173 * HISTORY
174 * November 12, 1997 - Andrew E. Mileski
175 * Written, tested, and released.
176 */
Cyrill Gorcunov28de7942007-07-21 04:37:18 -0700177static int udf_UTF8toCS0(dstring *ocu, struct ustr *utf, int length)
Linus Torvalds1da177e2005-04-16 15:20:36 -0700178{
179 unsigned c, i, max_val, utf_char;
180 int utf_cnt, u_len;
181
182 memset(ocu, 0, sizeof(dstring) * length);
183 ocu[0] = 8;
184 max_val = 0xffU;
185
Cyrill Gorcunov28de7942007-07-21 04:37:18 -0700186try_again:
Linus Torvalds1da177e2005-04-16 15:20:36 -0700187 u_len = 0U;
188 utf_char = 0U;
189 utf_cnt = 0U;
Cyrill Gorcunovcb00ea32007-07-19 01:47:43 -0700190 for (i = 0U; i < utf->u_len; i++) {
Cyrill Gorcunov28de7942007-07-21 04:37:18 -0700191 c = (uint8_t)utf->u_name[i];
Linus Torvalds1da177e2005-04-16 15:20:36 -0700192
193 /* Complete a multi-byte UTF-8 character */
Cyrill Gorcunovcb00ea32007-07-19 01:47:43 -0700194 if (utf_cnt) {
Linus Torvalds1da177e2005-04-16 15:20:36 -0700195 utf_char = (utf_char << 6) | (c & 0x3fU);
196 if (--utf_cnt)
197 continue;
Cyrill Gorcunovcb00ea32007-07-19 01:47:43 -0700198 } else {
Linus Torvalds1da177e2005-04-16 15:20:36 -0700199 /* Check for a multi-byte UTF-8 character */
Cyrill Gorcunovcb00ea32007-07-19 01:47:43 -0700200 if (c & 0x80U) {
Linus Torvalds1da177e2005-04-16 15:20:36 -0700201 /* Start a multi-byte UTF-8 character */
Cyrill Gorcunovcb00ea32007-07-19 01:47:43 -0700202 if ((c & 0xe0U) == 0xc0U) {
Linus Torvalds1da177e2005-04-16 15:20:36 -0700203 utf_char = c & 0x1fU;
204 utf_cnt = 1;
Cyrill Gorcunovcb00ea32007-07-19 01:47:43 -0700205 } else if ((c & 0xf0U) == 0xe0U) {
Linus Torvalds1da177e2005-04-16 15:20:36 -0700206 utf_char = c & 0x0fU;
207 utf_cnt = 2;
Cyrill Gorcunovcb00ea32007-07-19 01:47:43 -0700208 } else if ((c & 0xf8U) == 0xf0U) {
Linus Torvalds1da177e2005-04-16 15:20:36 -0700209 utf_char = c & 0x07U;
210 utf_cnt = 3;
Cyrill Gorcunovcb00ea32007-07-19 01:47:43 -0700211 } else if ((c & 0xfcU) == 0xf8U) {
Linus Torvalds1da177e2005-04-16 15:20:36 -0700212 utf_char = c & 0x03U;
213 utf_cnt = 4;
Cyrill Gorcunovcb00ea32007-07-19 01:47:43 -0700214 } else if ((c & 0xfeU) == 0xfcU) {
Linus Torvalds1da177e2005-04-16 15:20:36 -0700215 utf_char = c & 0x01U;
216 utf_cnt = 5;
Cyrill Gorcunov28de7942007-07-21 04:37:18 -0700217 } else {
Linus Torvalds1da177e2005-04-16 15:20:36 -0700218 goto error_out;
Cyrill Gorcunov28de7942007-07-21 04:37:18 -0700219 }
Linus Torvalds1da177e2005-04-16 15:20:36 -0700220 continue;
Cyrill Gorcunov28de7942007-07-21 04:37:18 -0700221 } else {
Linus Torvalds1da177e2005-04-16 15:20:36 -0700222 /* Single byte UTF-8 character (most common) */
223 utf_char = c;
Cyrill Gorcunov28de7942007-07-21 04:37:18 -0700224 }
Linus Torvalds1da177e2005-04-16 15:20:36 -0700225 }
226
227 /* Choose no compression if necessary */
Cyrill Gorcunovcb00ea32007-07-19 01:47:43 -0700228 if (utf_char > max_val) {
Cyrill Gorcunov28de7942007-07-21 04:37:18 -0700229 if (max_val == 0xffU) {
Linus Torvalds1da177e2005-04-16 15:20:36 -0700230 max_val = 0xffffU;
Cyrill Gorcunov28de7942007-07-21 04:37:18 -0700231 ocu[0] = (uint8_t)0x10U;
Linus Torvalds1da177e2005-04-16 15:20:36 -0700232 goto try_again;
233 }
234 goto error_out;
235 }
236
Marcin Slusarz4b111112008-02-08 04:20:36 -0800237 if (max_val == 0xffffU)
Cyrill Gorcunov28de7942007-07-21 04:37:18 -0700238 ocu[++u_len] = (uint8_t)(utf_char >> 8);
Cyrill Gorcunov28de7942007-07-21 04:37:18 -0700239 ocu[++u_len] = (uint8_t)(utf_char & 0xffU);
Linus Torvalds1da177e2005-04-16 15:20:36 -0700240 }
241
Cyrill Gorcunovcb00ea32007-07-19 01:47:43 -0700242 if (utf_cnt) {
Cyrill Gorcunov28de7942007-07-21 04:37:18 -0700243error_out:
Linus Torvalds1da177e2005-04-16 15:20:36 -0700244 ocu[++u_len] = '?';
Joe Perches78ace702011-10-10 01:08:05 -0700245 printk(KERN_DEBUG pr_fmt("bad UTF-8 character\n"));
Linus Torvalds1da177e2005-04-16 15:20:36 -0700246 }
247
Cyrill Gorcunov28de7942007-07-21 04:37:18 -0700248 ocu[length - 1] = (uint8_t)u_len + 1;
249
Linus Torvalds1da177e2005-04-16 15:20:36 -0700250 return u_len + 1;
251}
252
Cyrill Gorcunovcb00ea32007-07-19 01:47:43 -0700253static int udf_CS0toNLS(struct nls_table *nls, struct ustr *utf_o,
marcin.slusarz@gmail.com34f953d2008-02-27 22:38:36 +0100254 const struct ustr *ocu_i)
Linus Torvalds1da177e2005-04-16 15:20:36 -0700255{
marcin.slusarz@gmail.com34f953d2008-02-27 22:38:36 +0100256 const uint8_t *ocu;
Linus Torvalds1da177e2005-04-16 15:20:36 -0700257 uint8_t cmp_id, ocu_len;
Jan Kara59285c22009-02-04 19:46:11 +0100258 int i, len;
Linus Torvalds1da177e2005-04-16 15:20:36 -0700259
Linus Torvalds1da177e2005-04-16 15:20:36 -0700260
261 ocu_len = ocu_i->u_len;
Cyrill Gorcunovcb00ea32007-07-19 01:47:43 -0700262 if (ocu_len == 0) {
Linus Torvalds1da177e2005-04-16 15:20:36 -0700263 memset(utf_o, 0, sizeof(struct ustr));
Linus Torvalds1da177e2005-04-16 15:20:36 -0700264 return 0;
265 }
266
marcin.slusarz@gmail.com34f953d2008-02-27 22:38:36 +0100267 cmp_id = ocu_i->u_cmpID;
268 if (cmp_id != 8 && cmp_id != 16) {
269 memset(utf_o, 0, sizeof(struct ustr));
Joe Perches78ace702011-10-10 01:08:05 -0700270 pr_err("unknown compression code (%d) stri=%s\n",
Cyrill Gorcunovcb00ea32007-07-19 01:47:43 -0700271 cmp_id, ocu_i->u_name);
Linus Torvalds1da177e2005-04-16 15:20:36 -0700272 return 0;
273 }
274
marcin.slusarz@gmail.com34f953d2008-02-27 22:38:36 +0100275 ocu = ocu_i->u_name;
276 utf_o->u_len = 0;
Cyrill Gorcunovcb00ea32007-07-19 01:47:43 -0700277 for (i = 0; (i < ocu_len) && (utf_o->u_len <= (UDF_NAME_LEN - 3));) {
Linus Torvalds1da177e2005-04-16 15:20:36 -0700278 /* Expand OSTA compressed Unicode to Unicode */
marcin.slusarz@gmail.com34f953d2008-02-27 22:38:36 +0100279 uint32_t c = ocu[i++];
Linus Torvalds1da177e2005-04-16 15:20:36 -0700280 if (cmp_id == 16)
281 c = (c << 8) | ocu[i++];
282
Jan Kara59285c22009-02-04 19:46:11 +0100283 len = nls->uni2char(c, &utf_o->u_name[utf_o->u_len],
284 UDF_NAME_LEN - utf_o->u_len);
285 /* Valid character? */
286 if (len >= 0)
287 utf_o->u_len += len;
288 else
289 utf_o->u_name[utf_o->u_len++] = '?';
Linus Torvalds1da177e2005-04-16 15:20:36 -0700290 }
Cyrill Gorcunovcb00ea32007-07-19 01:47:43 -0700291 utf_o->u_cmpID = 8;
Linus Torvalds1da177e2005-04-16 15:20:36 -0700292
293 return utf_o->u_len;
294}
295
Cyrill Gorcunov28de7942007-07-21 04:37:18 -0700296static int udf_NLStoCS0(struct nls_table *nls, dstring *ocu, struct ustr *uni,
Cyrill Gorcunovcb00ea32007-07-19 01:47:43 -0700297 int length)
Linus Torvalds1da177e2005-04-16 15:20:36 -0700298{
Jan Kara59285c22009-02-04 19:46:11 +0100299 int len;
300 unsigned i, max_val;
Linus Torvalds1da177e2005-04-16 15:20:36 -0700301 uint16_t uni_char;
302 int u_len;
303
304 memset(ocu, 0, sizeof(dstring) * length);
305 ocu[0] = 8;
306 max_val = 0xffU;
307
Cyrill Gorcunov28de7942007-07-21 04:37:18 -0700308try_again:
Linus Torvalds1da177e2005-04-16 15:20:36 -0700309 u_len = 0U;
Cyrill Gorcunovcb00ea32007-07-19 01:47:43 -0700310 for (i = 0U; i < uni->u_len; i++) {
311 len = nls->char2uni(&uni->u_name[i], uni->u_len - i, &uni_char);
Jan Kara59285c22009-02-04 19:46:11 +0100312 if (!len)
Linus Torvalds1da177e2005-04-16 15:20:36 -0700313 continue;
Jan Kara59285c22009-02-04 19:46:11 +0100314 /* Invalid character, deal with it */
315 if (len < 0) {
316 len = 1;
317 uni_char = '?';
318 }
Linus Torvalds1da177e2005-04-16 15:20:36 -0700319
Cyrill Gorcunovcb00ea32007-07-19 01:47:43 -0700320 if (uni_char > max_val) {
Linus Torvalds1da177e2005-04-16 15:20:36 -0700321 max_val = 0xffffU;
Cyrill Gorcunov28de7942007-07-21 04:37:18 -0700322 ocu[0] = (uint8_t)0x10U;
Linus Torvalds1da177e2005-04-16 15:20:36 -0700323 goto try_again;
324 }
Cyrill Gorcunovcb00ea32007-07-19 01:47:43 -0700325
Linus Torvalds1da177e2005-04-16 15:20:36 -0700326 if (max_val == 0xffffU)
Cyrill Gorcunov28de7942007-07-21 04:37:18 -0700327 ocu[++u_len] = (uint8_t)(uni_char >> 8);
328 ocu[++u_len] = (uint8_t)(uni_char & 0xffU);
Linus Torvalds1da177e2005-04-16 15:20:36 -0700329 i += len - 1;
330 }
331
Cyrill Gorcunov28de7942007-07-21 04:37:18 -0700332 ocu[length - 1] = (uint8_t)u_len + 1;
Linus Torvalds1da177e2005-04-16 15:20:36 -0700333 return u_len + 1;
334}
335
Cyrill Gorcunov28de7942007-07-21 04:37:18 -0700336int udf_get_filename(struct super_block *sb, uint8_t *sname, uint8_t *dname,
Cyrill Gorcunovcb00ea32007-07-19 01:47:43 -0700337 int flen)
Linus Torvalds1da177e2005-04-16 15:20:36 -0700338{
Marcin Slusarz530f1a52008-11-16 19:02:45 +0100339 struct ustr *filename, *unifilename;
340 int len = 0;
Linus Torvalds1da177e2005-04-16 15:20:36 -0700341
Marcin Slusarz530f1a52008-11-16 19:02:45 +0100342 filename = kmalloc(sizeof(struct ustr), GFP_NOFS);
343 if (!filename)
Linus Torvalds1da177e2005-04-16 15:20:36 -0700344 return 0;
Linus Torvalds1da177e2005-04-16 15:20:36 -0700345
Marcin Slusarz530f1a52008-11-16 19:02:45 +0100346 unifilename = kmalloc(sizeof(struct ustr), GFP_NOFS);
347 if (!unifilename)
348 goto out1;
349
350 if (udf_build_ustr_exact(unifilename, sname, flen))
351 goto out2;
352
Cyrill Gorcunovcb00ea32007-07-19 01:47:43 -0700353 if (UDF_QUERY_FLAG(sb, UDF_FLAG_UTF8)) {
Marcin Slusarz530f1a52008-11-16 19:02:45 +0100354 if (!udf_CS0toUTF8(filename, unifilename)) {
Marcin Slusarz4b111112008-02-08 04:20:36 -0800355 udf_debug("Failed in udf_get_filename: sname = %s\n",
356 sname);
Marcin Slusarz530f1a52008-11-16 19:02:45 +0100357 goto out2;
Linus Torvalds1da177e2005-04-16 15:20:36 -0700358 }
Cyrill Gorcunovcb00ea32007-07-19 01:47:43 -0700359 } else if (UDF_QUERY_FLAG(sb, UDF_FLAG_NLS_MAP)) {
Marcin Slusarz530f1a52008-11-16 19:02:45 +0100360 if (!udf_CS0toNLS(UDF_SB(sb)->s_nls_map, filename,
361 unifilename)) {
Marcin Slusarz4b111112008-02-08 04:20:36 -0800362 udf_debug("Failed in udf_get_filename: sname = %s\n",
363 sname);
Marcin Slusarz530f1a52008-11-16 19:02:45 +0100364 goto out2;
Linus Torvalds1da177e2005-04-16 15:20:36 -0700365 }
Marcin Slusarz4b111112008-02-08 04:20:36 -0800366 } else
Marcin Slusarz530f1a52008-11-16 19:02:45 +0100367 goto out2;
Linus Torvalds1da177e2005-04-16 15:20:36 -0700368
Marcin Slusarz530f1a52008-11-16 19:02:45 +0100369 len = udf_translate_to_linux(dname, filename->u_name, filename->u_len,
370 unifilename->u_name, unifilename->u_len);
371out2:
372 kfree(unifilename);
373out1:
374 kfree(filename);
375 return len;
Linus Torvalds1da177e2005-04-16 15:20:36 -0700376}
377
Cyrill Gorcunov28de7942007-07-21 04:37:18 -0700378int udf_put_filename(struct super_block *sb, const uint8_t *sname,
379 uint8_t *dname, int flen)
Linus Torvalds1da177e2005-04-16 15:20:36 -0700380{
381 struct ustr unifilename;
382 int namelen;
383
Marcin Slusarz4b111112008-02-08 04:20:36 -0800384 if (!udf_char_to_ustr(&unifilename, sname, flen))
Linus Torvalds1da177e2005-04-16 15:20:36 -0700385 return 0;
Linus Torvalds1da177e2005-04-16 15:20:36 -0700386
Cyrill Gorcunovcb00ea32007-07-19 01:47:43 -0700387 if (UDF_QUERY_FLAG(sb, UDF_FLAG_UTF8)) {
Cyrill Gorcunov28de7942007-07-21 04:37:18 -0700388 namelen = udf_UTF8toCS0(dname, &unifilename, UDF_NAME_LEN);
Marcin Slusarz4b111112008-02-08 04:20:36 -0800389 if (!namelen)
Linus Torvalds1da177e2005-04-16 15:20:36 -0700390 return 0;
Cyrill Gorcunovcb00ea32007-07-19 01:47:43 -0700391 } else if (UDF_QUERY_FLAG(sb, UDF_FLAG_NLS_MAP)) {
Marcin Slusarz4b111112008-02-08 04:20:36 -0800392 namelen = udf_NLStoCS0(UDF_SB(sb)->s_nls_map, dname,
393 &unifilename, UDF_NAME_LEN);
394 if (!namelen)
Linus Torvalds1da177e2005-04-16 15:20:36 -0700395 return 0;
Marcin Slusarz4b111112008-02-08 04:20:36 -0800396 } else
Linus Torvalds1da177e2005-04-16 15:20:36 -0700397 return 0;
398
399 return namelen;
400}
401
402#define ILLEGAL_CHAR_MARK '_'
Cyrill Gorcunov28de7942007-07-21 04:37:18 -0700403#define EXT_MARK '.'
404#define CRC_MARK '#'
405#define EXT_SIZE 5
Linus Torvalds1da177e2005-04-16 15:20:36 -0700406
Marcin Slusarz4b111112008-02-08 04:20:36 -0800407static int udf_translate_to_linux(uint8_t *newName, uint8_t *udfName,
408 int udfLen, uint8_t *fidName,
409 int fidNameLen)
Linus Torvalds1da177e2005-04-16 15:20:36 -0700410{
Cyrill Gorcunovcb00ea32007-07-19 01:47:43 -0700411 int index, newIndex = 0, needsCRC = 0;
Linus Torvalds1da177e2005-04-16 15:20:36 -0700412 int extIndex = 0, newExtIndex = 0, hasExt = 0;
413 unsigned short valueCRC;
414 uint8_t curr;
415 const uint8_t hexChar[] = "0123456789ABCDEF";
416
Cyrill Gorcunov28de7942007-07-21 04:37:18 -0700417 if (udfName[0] == '.' &&
418 (udfLen == 1 || (udfLen == 2 && udfName[1] == '.'))) {
Linus Torvalds1da177e2005-04-16 15:20:36 -0700419 needsCRC = 1;
420 newIndex = udfLen;
421 memcpy(newName, udfName, udfLen);
Cyrill Gorcunovcb00ea32007-07-19 01:47:43 -0700422 } else {
423 for (index = 0; index < udfLen; index++) {
Linus Torvalds1da177e2005-04-16 15:20:36 -0700424 curr = udfName[index];
Cyrill Gorcunovcb00ea32007-07-19 01:47:43 -0700425 if (curr == '/' || curr == 0) {
Linus Torvalds1da177e2005-04-16 15:20:36 -0700426 needsCRC = 1;
427 curr = ILLEGAL_CHAR_MARK;
Marcin Slusarz4b111112008-02-08 04:20:36 -0800428 while (index + 1 < udfLen &&
429 (udfName[index + 1] == '/' ||
430 udfName[index + 1] == 0))
Linus Torvalds1da177e2005-04-16 15:20:36 -0700431 index++;
Marcin Slusarz4b111112008-02-08 04:20:36 -0800432 }
433 if (curr == EXT_MARK &&
434 (udfLen - index - 1) <= EXT_SIZE) {
435 if (udfLen == index + 1)
Linus Torvalds1da177e2005-04-16 15:20:36 -0700436 hasExt = 0;
Marcin Slusarz4b111112008-02-08 04:20:36 -0800437 else {
Linus Torvalds1da177e2005-04-16 15:20:36 -0700438 hasExt = 1;
439 extIndex = index;
440 newExtIndex = newIndex;
441 }
442 }
443 if (newIndex < 256)
444 newName[newIndex++] = curr;
445 else
446 needsCRC = 1;
447 }
448 }
Cyrill Gorcunovcb00ea32007-07-19 01:47:43 -0700449 if (needsCRC) {
Linus Torvalds1da177e2005-04-16 15:20:36 -0700450 uint8_t ext[EXT_SIZE];
451 int localExtIndex = 0;
452
Cyrill Gorcunovcb00ea32007-07-19 01:47:43 -0700453 if (hasExt) {
Linus Torvalds1da177e2005-04-16 15:20:36 -0700454 int maxFilenameLen;
Marcin Slusarz4b111112008-02-08 04:20:36 -0800455 for (index = 0;
456 index < EXT_SIZE && extIndex + index + 1 < udfLen;
457 index++) {
Linus Torvalds1da177e2005-04-16 15:20:36 -0700458 curr = udfName[extIndex + index + 1];
459
Cyrill Gorcunovcb00ea32007-07-19 01:47:43 -0700460 if (curr == '/' || curr == 0) {
Linus Torvalds1da177e2005-04-16 15:20:36 -0700461 needsCRC = 1;
462 curr = ILLEGAL_CHAR_MARK;
Marcin Slusarz4b111112008-02-08 04:20:36 -0800463 while (extIndex + index + 2 < udfLen &&
464 (index + 1 < EXT_SIZE &&
465 (udfName[extIndex + index + 2] == '/' ||
466 udfName[extIndex + index + 2] == 0)))
Linus Torvalds1da177e2005-04-16 15:20:36 -0700467 index++;
468 }
469 ext[localExtIndex++] = curr;
470 }
471 maxFilenameLen = 250 - localExtIndex;
472 if (newIndex > maxFilenameLen)
473 newIndex = maxFilenameLen;
474 else
475 newIndex = newExtIndex;
Marcin Slusarz4b111112008-02-08 04:20:36 -0800476 } else if (newIndex > 250)
Linus Torvalds1da177e2005-04-16 15:20:36 -0700477 newIndex = 250;
478 newName[newIndex++] = CRC_MARK;
Bob Copelandf845fce2008-04-17 09:47:48 +0200479 valueCRC = crc_itu_t(0, fidName, fidNameLen);
Linus Torvalds1da177e2005-04-16 15:20:36 -0700480 newName[newIndex++] = hexChar[(valueCRC & 0xf000) >> 12];
481 newName[newIndex++] = hexChar[(valueCRC & 0x0f00) >> 8];
482 newName[newIndex++] = hexChar[(valueCRC & 0x00f0) >> 4];
483 newName[newIndex++] = hexChar[(valueCRC & 0x000f)];
484
Cyrill Gorcunovcb00ea32007-07-19 01:47:43 -0700485 if (hasExt) {
Linus Torvalds1da177e2005-04-16 15:20:36 -0700486 newName[newIndex++] = EXT_MARK;
Cyrill Gorcunovcb00ea32007-07-19 01:47:43 -0700487 for (index = 0; index < localExtIndex; index++)
Linus Torvalds1da177e2005-04-16 15:20:36 -0700488 newName[newIndex++] = ext[index];
489 }
490 }
Cyrill Gorcunov28de7942007-07-21 04:37:18 -0700491
Linus Torvalds1da177e2005-04-16 15:20:36 -0700492 return newIndex;
493}