5 * Routines for converting between UTF-8 and OSTA Compressed Unicode.
6 * Also handles filename mangling
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
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.
23 #include <linux/kernel.h>
24 #include <linux/string.h> /* for memset */
25 #include <linux/nls.h>
26 #include <linux/crc-itu-t.h>
27 #include <linux/slab.h>
31 #define PLANE_SIZE 0x10000
32 #define UNICODE_MAX 0x10ffff
33 #define SURROGATE_MASK 0xfffff800
34 #define SURROGATE_PAIR 0x0000d800
35 #define SURROGATE_LOW 0x00000400
36 #define SURROGATE_CHAR_BITS 10
37 #define SURROGATE_CHAR_MASK ((1 << SURROGATE_CHAR_BITS) - 1)
39 #define ILLEGAL_CHAR_MARK '_'
43 /* Number of chars we need to store generated CRC to make filename unique */
46 static unicode_t get_utf16_char(const uint8_t *str_i, int str_i_max_len,
47 int str_i_idx, int u_ch, unicode_t *ret)
50 int start_idx = str_i_idx;
52 /* Expand OSTA compressed Unicode to Unicode */
53 c = str_i[str_i_idx++];
55 c = (c << 8) | str_i[str_i_idx++];
56 if ((c & SURROGATE_MASK) == SURROGATE_PAIR) {
59 /* Trailing surrogate char */
60 if (str_i_idx >= str_i_max_len) {
65 /* Low surrogate must follow the high one... */
66 if (c & SURROGATE_LOW) {
71 WARN_ON_ONCE(u_ch != 2);
72 next = str_i[str_i_idx++] << 8;
73 next |= str_i[str_i_idx++];
74 if ((next & SURROGATE_MASK) != SURROGATE_PAIR ||
75 !(next & SURROGATE_LOW)) {
81 ((c & SURROGATE_CHAR_MASK) << SURROGATE_CHAR_BITS) +
82 (next & SURROGATE_CHAR_MASK);
86 return str_i_idx - start_idx;
90 static int udf_name_conv_char(uint8_t *str_o, int str_o_max_len,
92 const uint8_t *str_i, int str_i_max_len,
94 int u_ch, int *needsCRC,
95 int (*conv_f)(wchar_t, unsigned char *, int),
102 while (!gotch && *str_i_idx < str_i_max_len) {
103 if (*str_o_idx >= str_o_max_len) {
108 len = get_utf16_char(str_i, str_i_max_len, *str_i_idx, u_ch,
110 /* These chars cannot be converted. Replace them. */
111 if (c == 0 || c > UNICODE_MAX || (conv_f && c > MAX_WCHAR_T) ||
112 (translate && c == '/')) {
124 c = ILLEGAL_CHAR_MARK;
129 len = conv_f(c, &str_o[*str_o_idx],
130 str_o_max_len - *str_o_idx);
132 len = utf32_to_utf8(c, &str_o[*str_o_idx],
133 str_o_max_len - *str_o_idx);
137 /* Valid character? */
140 else if (len == -ENAMETOOLONG) {
144 str_o[(*str_o_idx)++] = ILLEGAL_CHAR_MARK;
151 static int udf_name_from_CS0(struct super_block *sb,
152 uint8_t *str_o, int str_max_len,
153 const uint8_t *ocu, int ocu_len,
161 int ext_i_len, ext_max_len;
162 int str_o_len = 0; /* Length of resulting output */
163 int ext_o_len = 0; /* Extension output length */
164 int ext_crc_len = 0; /* Extension output length if used with CRC */
165 int i_ext = -1; /* Extension position in input buffer */
166 int o_crc = 0; /* Rightmost possible output pos for CRC+ext */
167 unsigned short valueCRC;
168 uint8_t ext[EXT_SIZE * NLS_MAX_CHARSET_SIZE + 1];
169 uint8_t crc[CRC_LEN];
170 int (*conv_f)(wchar_t, unsigned char *, int);
172 if (str_max_len <= 0)
176 memset(str_o, 0, str_max_len);
180 if (UDF_SB(sb)->s_nls_map)
181 conv_f = UDF_SB(sb)->s_nls_map->uni2char;
186 if (cmp_id != 8 && cmp_id != 16) {
187 memset(str_o, 0, str_max_len);
188 pr_err("unknown compression code (%u)\n", cmp_id);
196 if (ocu_len % u_ch) {
197 pr_err("incorrect filename length (%d)\n", ocu_len + 1);
202 /* Look for extension */
203 for (idx = ocu_len - u_ch, ext_i_len = 0;
204 (idx >= 0) && (ext_i_len < EXT_SIZE);
205 idx -= u_ch, ext_i_len++) {
208 c = (c << 8) | ocu[idx + 1];
217 /* Convert extension */
218 ext_max_len = min_t(int, sizeof(ext), str_max_len);
219 ext[ext_o_len++] = EXT_MARK;
221 while (udf_name_conv_char(ext, ext_max_len, &ext_o_len,
224 conv_f, translate)) {
225 if ((ext_o_len + CRC_LEN) < str_max_len)
226 ext_crc_len = ext_o_len;
233 if (translate && (idx == i_ext)) {
234 if (str_o_len > (str_max_len - ext_o_len))
239 if (!udf_name_conv_char(str_o, str_max_len, &str_o_len,
241 u_ch, &needsCRC, conv_f, translate))
245 (str_o_len <= (str_max_len - ext_o_len - CRC_LEN)))
250 if (str_o_len > 0 && str_o_len <= 2 && str_o[0] == '.' &&
251 (str_o_len == 1 || str_o[1] == '.'))
255 valueCRC = crc_itu_t(0, ocu, ocu_len);
257 crc[1] = hex_asc_upper_hi(valueCRC >> 8);
258 crc[2] = hex_asc_upper_lo(valueCRC >> 8);
259 crc[3] = hex_asc_upper_hi(valueCRC);
260 crc[4] = hex_asc_upper_lo(valueCRC);
261 len = min_t(int, CRC_LEN, str_max_len - str_o_len);
262 memcpy(&str_o[str_o_len], crc, len);
264 ext_o_len = ext_crc_len;
267 memcpy(&str_o[str_o_len], ext, ext_o_len);
268 str_o_len += ext_o_len;
275 static int udf_name_to_CS0(struct super_block *sb,
276 uint8_t *ocu, int ocu_max_len,
277 const uint8_t *str_i, int str_len)
280 unsigned int max_val;
283 int (*conv_f)(const unsigned char *, int, wchar_t *);
285 if (ocu_max_len <= 0)
288 if (UDF_SB(sb)->s_nls_map)
289 conv_f = UDF_SB(sb)->s_nls_map->char2uni;
293 memset(ocu, 0, ocu_max_len);
300 for (i = 0; i < str_len; i += len) {
301 /* Name didn't fit? */
302 if (u_len + u_ch > ocu_max_len)
307 len = conv_f(&str_i[i], str_len - i, &wchar);
311 len = utf8_to_utf32(&str_i[i], str_len - i,
314 /* Invalid character, deal with it */
315 if (len <= 0 || uni_char > UNICODE_MAX) {
320 if (uni_char > max_val) {
323 if (max_val == 0xff) {
330 * Use UTF-16 encoding for chars outside we
331 * cannot encode directly.
333 if (u_len + 2 * u_ch > ocu_max_len)
336 uni_char -= PLANE_SIZE;
338 ((uni_char >> SURROGATE_CHAR_BITS) &
339 SURROGATE_CHAR_MASK);
340 ocu[u_len++] = (uint8_t)(c >> 8);
341 ocu[u_len++] = (uint8_t)(c & 0xff);
342 uni_char = SURROGATE_PAIR | SURROGATE_LOW |
343 (uni_char & SURROGATE_CHAR_MASK);
346 if (max_val == 0xffff)
347 ocu[u_len++] = (uint8_t)(uni_char >> 8);
348 ocu[u_len++] = (uint8_t)(uni_char & 0xff);
355 * Convert CS0 dstring to output charset. Warning: This function may truncate
356 * input string if it is too long as it is used for informational strings only
357 * and it is better to truncate the string than to refuse mounting a media.
359 int udf_dstrCS0toChar(struct super_block *sb, uint8_t *utf_o, int o_len,
360 const uint8_t *ocu_i, int i_len)
365 s_len = ocu_i[i_len - 1];
366 if (s_len >= i_len) {
367 pr_warn("incorrect dstring lengths (%d/%d),"
368 " truncating\n", s_len, i_len);
370 /* 2-byte encoding? Need to round properly... */
372 s_len -= (s_len - 1) & 2;
376 return udf_name_from_CS0(sb, utf_o, o_len, ocu_i, s_len, 0);
379 int udf_get_filename(struct super_block *sb, const uint8_t *sname, int slen,
380 uint8_t *dname, int dlen)
390 ret = udf_name_from_CS0(sb, dname, dlen, sname, slen, 1);
391 /* Zero length filename isn't valid... */
397 int udf_put_filename(struct super_block *sb, const uint8_t *sname, int slen,
398 uint8_t *dname, int dlen)
400 return udf_name_to_CS0(sb, dname, dlen, sname, slen);