]>
Commit | Line | Data |
---|---|---|
1da177e4 LT |
1 | /* |
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 | * | |
1da177e4 LT |
14 | * 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> | |
1da177e4 LT |
26 | |
27 | #include "udf_sb.h" | |
28 | ||
29 | static int udf_translate_to_linux(uint8_t *, uint8_t *, int, uint8_t *, int); | |
30 | ||
28de7948 | 31 | static int udf_char_to_ustr(struct ustr *dest, const uint8_t *src, int strlen) |
1da177e4 | 32 | { |
cb00ea35 | 33 | if ((!dest) || (!src) || (!strlen) || (strlen > UDF_NAME_LEN - 2)) |
1da177e4 | 34 | return 0; |
28de7948 | 35 | |
1da177e4 LT |
36 | memset(dest, 0, sizeof(struct ustr)); |
37 | memcpy(dest->u_name, src, strlen); | |
38 | dest->u_cmpID = 0x08; | |
39 | dest->u_len = strlen; | |
28de7948 | 40 | |
1da177e4 LT |
41 | return strlen; |
42 | } | |
43 | ||
44 | /* | |
45 | * udf_build_ustr | |
46 | */ | |
28de7948 | 47 | int udf_build_ustr(struct ustr *dest, dstring *ptr, int size) |
1da177e4 LT |
48 | { |
49 | int usesize; | |
50 | ||
6305a0a9 | 51 | if (!dest || !ptr || !size) |
1da177e4 | 52 | return -1; |
6305a0a9 | 53 | BUG_ON(size < 2); |
1da177e4 | 54 | |
6305a0a9 MS |
55 | usesize = min_t(size_t, ptr[size - 1], sizeof(dest->u_name)); |
56 | usesize = min(usesize, size - 2); | |
cb00ea35 | 57 | dest->u_cmpID = ptr[0]; |
6305a0a9 MS |
58 | dest->u_len = usesize; |
59 | memcpy(dest->u_name, ptr + 1, usesize); | |
60 | memset(dest->u_name + usesize, 0, sizeof(dest->u_name) - usesize); | |
28de7948 | 61 | |
1da177e4 LT |
62 | return 0; |
63 | } | |
64 | ||
65 | /* | |
66 | * udf_build_ustr_exact | |
67 | */ | |
28de7948 | 68 | static int udf_build_ustr_exact(struct ustr *dest, dstring *ptr, int exactsize) |
1da177e4 | 69 | { |
cb00ea35 | 70 | if ((!dest) || (!ptr) || (!exactsize)) |
1da177e4 LT |
71 | return -1; |
72 | ||
73 | memset(dest, 0, sizeof(struct ustr)); | |
cb00ea35 CG |
74 | dest->u_cmpID = ptr[0]; |
75 | dest->u_len = exactsize - 1; | |
76 | memcpy(dest->u_name, ptr + 1, exactsize - 1); | |
28de7948 | 77 | |
1da177e4 LT |
78 | return 0; |
79 | } | |
80 | ||
81 | /* | |
82 | * udf_ocu_to_utf8 | |
83 | * | |
84 | * PURPOSE | |
85 | * Convert OSTA Compressed Unicode to the UTF-8 equivalent. | |
86 | * | |
1da177e4 LT |
87 | * PRE-CONDITIONS |
88 | * utf Pointer to UTF-8 output buffer. | |
89 | * ocu Pointer to OSTA Compressed Unicode input buffer | |
90 | * of size UDF_NAME_LEN bytes. | |
91 | * both of type "struct ustr *" | |
92 | * | |
93 | * POST-CONDITIONS | |
94 | * <return> Zero on success. | |
95 | * | |
96 | * HISTORY | |
97 | * November 12, 1997 - Andrew E. Mileski | |
98 | * Written, tested, and released. | |
99 | */ | |
79cfe0ff | 100 | int udf_CS0toUTF8(struct ustr *utf_o, const struct ustr *ocu_i) |
1da177e4 | 101 | { |
79cfe0ff | 102 | const uint8_t *ocu; |
1da177e4 LT |
103 | uint8_t cmp_id, ocu_len; |
104 | int i; | |
105 | ||
1da177e4 | 106 | ocu_len = ocu_i->u_len; |
cb00ea35 | 107 | if (ocu_len == 0) { |
1da177e4 | 108 | memset(utf_o, 0, sizeof(struct ustr)); |
1da177e4 LT |
109 | return 0; |
110 | } | |
111 | ||
79cfe0ff | 112 | cmp_id = ocu_i->u_cmpID; |
113 | if (cmp_id != 8 && cmp_id != 16) { | |
114 | memset(utf_o, 0, sizeof(struct ustr)); | |
cb00ea35 CG |
115 | printk(KERN_ERR "udf: unknown compression code (%d) stri=%s\n", |
116 | cmp_id, ocu_i->u_name); | |
1da177e4 LT |
117 | return 0; |
118 | } | |
119 | ||
79cfe0ff | 120 | ocu = ocu_i->u_name; |
121 | utf_o->u_len = 0; | |
cb00ea35 | 122 | for (i = 0; (i < ocu_len) && (utf_o->u_len <= (UDF_NAME_LEN - 3));) { |
1da177e4 LT |
123 | |
124 | /* Expand OSTA compressed Unicode to Unicode */ | |
79cfe0ff | 125 | uint32_t c = ocu[i++]; |
1da177e4 LT |
126 | if (cmp_id == 16) |
127 | c = (c << 8) | ocu[i++]; | |
128 | ||
129 | /* Compress Unicode to UTF-8 */ | |
79cfe0ff | 130 | if (c < 0x80U) |
28de7948 | 131 | utf_o->u_name[utf_o->u_len++] = (uint8_t)c; |
79cfe0ff | 132 | else if (c < 0x800U) { |
4b11111a MS |
133 | utf_o->u_name[utf_o->u_len++] = |
134 | (uint8_t)(0xc0 | (c >> 6)); | |
135 | utf_o->u_name[utf_o->u_len++] = | |
136 | (uint8_t)(0x80 | (c & 0x3f)); | |
cb00ea35 | 137 | } else { |
4b11111a MS |
138 | utf_o->u_name[utf_o->u_len++] = |
139 | (uint8_t)(0xe0 | (c >> 12)); | |
140 | utf_o->u_name[utf_o->u_len++] = | |
141 | (uint8_t)(0x80 | | |
142 | ((c >> 6) & 0x3f)); | |
143 | utf_o->u_name[utf_o->u_len++] = | |
144 | (uint8_t)(0x80 | (c & 0x3f)); | |
1da177e4 LT |
145 | } |
146 | } | |
cb00ea35 | 147 | utf_o->u_cmpID = 8; |
1da177e4 LT |
148 | |
149 | return utf_o->u_len; | |
150 | } | |
151 | ||
152 | /* | |
153 | * | |
154 | * udf_utf8_to_ocu | |
155 | * | |
156 | * PURPOSE | |
157 | * Convert UTF-8 to the OSTA Compressed Unicode equivalent. | |
158 | * | |
159 | * DESCRIPTION | |
160 | * This routine is only called by udf_lookup(). | |
161 | * | |
162 | * PRE-CONDITIONS | |
163 | * ocu Pointer to OSTA Compressed Unicode output | |
164 | * buffer of size UDF_NAME_LEN bytes. | |
165 | * utf Pointer to UTF-8 input buffer. | |
166 | * utf_len Length of UTF-8 input buffer in bytes. | |
167 | * | |
168 | * POST-CONDITIONS | |
169 | * <return> Zero on success. | |
170 | * | |
171 | * HISTORY | |
172 | * November 12, 1997 - Andrew E. Mileski | |
173 | * Written, tested, and released. | |
174 | */ | |
28de7948 | 175 | static int udf_UTF8toCS0(dstring *ocu, struct ustr *utf, int length) |
1da177e4 LT |
176 | { |
177 | unsigned c, i, max_val, utf_char; | |
178 | int utf_cnt, u_len; | |
179 | ||
180 | memset(ocu, 0, sizeof(dstring) * length); | |
181 | ocu[0] = 8; | |
182 | max_val = 0xffU; | |
183 | ||
28de7948 | 184 | try_again: |
1da177e4 LT |
185 | u_len = 0U; |
186 | utf_char = 0U; | |
187 | utf_cnt = 0U; | |
cb00ea35 | 188 | for (i = 0U; i < utf->u_len; i++) { |
28de7948 | 189 | c = (uint8_t)utf->u_name[i]; |
1da177e4 LT |
190 | |
191 | /* Complete a multi-byte UTF-8 character */ | |
cb00ea35 | 192 | if (utf_cnt) { |
1da177e4 LT |
193 | utf_char = (utf_char << 6) | (c & 0x3fU); |
194 | if (--utf_cnt) | |
195 | continue; | |
cb00ea35 | 196 | } else { |
1da177e4 | 197 | /* Check for a multi-byte UTF-8 character */ |
cb00ea35 | 198 | if (c & 0x80U) { |
1da177e4 | 199 | /* Start a multi-byte UTF-8 character */ |
cb00ea35 | 200 | if ((c & 0xe0U) == 0xc0U) { |
1da177e4 LT |
201 | utf_char = c & 0x1fU; |
202 | utf_cnt = 1; | |
cb00ea35 | 203 | } else if ((c & 0xf0U) == 0xe0U) { |
1da177e4 LT |
204 | utf_char = c & 0x0fU; |
205 | utf_cnt = 2; | |
cb00ea35 | 206 | } else if ((c & 0xf8U) == 0xf0U) { |
1da177e4 LT |
207 | utf_char = c & 0x07U; |
208 | utf_cnt = 3; | |
cb00ea35 | 209 | } else if ((c & 0xfcU) == 0xf8U) { |
1da177e4 LT |
210 | utf_char = c & 0x03U; |
211 | utf_cnt = 4; | |
cb00ea35 | 212 | } else if ((c & 0xfeU) == 0xfcU) { |
1da177e4 LT |
213 | utf_char = c & 0x01U; |
214 | utf_cnt = 5; | |
28de7948 | 215 | } else { |
1da177e4 | 216 | goto error_out; |
28de7948 | 217 | } |
1da177e4 | 218 | continue; |
28de7948 | 219 | } else { |
1da177e4 LT |
220 | /* Single byte UTF-8 character (most common) */ |
221 | utf_char = c; | |
28de7948 | 222 | } |
1da177e4 LT |
223 | } |
224 | ||
225 | /* Choose no compression if necessary */ | |
cb00ea35 | 226 | if (utf_char > max_val) { |
28de7948 | 227 | if (max_val == 0xffU) { |
1da177e4 | 228 | max_val = 0xffffU; |
28de7948 | 229 | ocu[0] = (uint8_t)0x10U; |
1da177e4 LT |
230 | goto try_again; |
231 | } | |
232 | goto error_out; | |
233 | } | |
234 | ||
4b11111a | 235 | if (max_val == 0xffffU) |
28de7948 | 236 | ocu[++u_len] = (uint8_t)(utf_char >> 8); |
28de7948 | 237 | ocu[++u_len] = (uint8_t)(utf_char & 0xffU); |
1da177e4 LT |
238 | } |
239 | ||
cb00ea35 | 240 | if (utf_cnt) { |
28de7948 | 241 | error_out: |
1da177e4 LT |
242 | ocu[++u_len] = '?'; |
243 | printk(KERN_DEBUG "udf: bad UTF-8 character\n"); | |
244 | } | |
245 | ||
28de7948 CG |
246 | ocu[length - 1] = (uint8_t)u_len + 1; |
247 | ||
1da177e4 LT |
248 | return u_len + 1; |
249 | } | |
250 | ||
cb00ea35 CG |
251 | static int udf_CS0toNLS(struct nls_table *nls, struct ustr *utf_o, |
252 | struct ustr *ocu_i) | |
1da177e4 LT |
253 | { |
254 | uint8_t *ocu; | |
255 | uint32_t c; | |
256 | uint8_t cmp_id, ocu_len; | |
257 | int i; | |
258 | ||
259 | ocu = ocu_i->u_name; | |
260 | ||
261 | ocu_len = ocu_i->u_len; | |
262 | cmp_id = ocu_i->u_cmpID; | |
263 | utf_o->u_len = 0; | |
264 | ||
cb00ea35 | 265 | if (ocu_len == 0) { |
1da177e4 LT |
266 | memset(utf_o, 0, sizeof(struct ustr)); |
267 | utf_o->u_cmpID = 0; | |
268 | utf_o->u_len = 0; | |
269 | return 0; | |
270 | } | |
271 | ||
cb00ea35 CG |
272 | if ((cmp_id != 8) && (cmp_id != 16)) { |
273 | printk(KERN_ERR "udf: unknown compression code (%d) stri=%s\n", | |
274 | cmp_id, ocu_i->u_name); | |
1da177e4 LT |
275 | return 0; |
276 | } | |
277 | ||
cb00ea35 | 278 | for (i = 0; (i < ocu_len) && (utf_o->u_len <= (UDF_NAME_LEN - 3));) { |
1da177e4 LT |
279 | /* Expand OSTA compressed Unicode to Unicode */ |
280 | c = ocu[i++]; | |
281 | if (cmp_id == 16) | |
282 | c = (c << 8) | ocu[i++]; | |
283 | ||
cb00ea35 CG |
284 | utf_o->u_len += nls->uni2char(c, &utf_o->u_name[utf_o->u_len], |
285 | UDF_NAME_LEN - utf_o->u_len); | |
1da177e4 | 286 | } |
cb00ea35 | 287 | utf_o->u_cmpID = 8; |
1da177e4 LT |
288 | |
289 | return utf_o->u_len; | |
290 | } | |
291 | ||
28de7948 | 292 | static int udf_NLStoCS0(struct nls_table *nls, dstring *ocu, struct ustr *uni, |
cb00ea35 | 293 | int length) |
1da177e4 LT |
294 | { |
295 | unsigned len, i, max_val; | |
296 | uint16_t uni_char; | |
297 | int u_len; | |
298 | ||
299 | memset(ocu, 0, sizeof(dstring) * length); | |
300 | ocu[0] = 8; | |
301 | max_val = 0xffU; | |
302 | ||
28de7948 | 303 | try_again: |
1da177e4 | 304 | u_len = 0U; |
cb00ea35 CG |
305 | for (i = 0U; i < uni->u_len; i++) { |
306 | len = nls->char2uni(&uni->u_name[i], uni->u_len - i, &uni_char); | |
1da177e4 LT |
307 | if (len <= 0) |
308 | continue; | |
309 | ||
cb00ea35 | 310 | if (uni_char > max_val) { |
1da177e4 | 311 | max_val = 0xffffU; |
28de7948 | 312 | ocu[0] = (uint8_t)0x10U; |
1da177e4 LT |
313 | goto try_again; |
314 | } | |
cb00ea35 | 315 | |
1da177e4 | 316 | if (max_val == 0xffffU) |
28de7948 CG |
317 | ocu[++u_len] = (uint8_t)(uni_char >> 8); |
318 | ocu[++u_len] = (uint8_t)(uni_char & 0xffU); | |
1da177e4 LT |
319 | i += len - 1; |
320 | } | |
321 | ||
28de7948 | 322 | ocu[length - 1] = (uint8_t)u_len + 1; |
1da177e4 LT |
323 | return u_len + 1; |
324 | } | |
325 | ||
28de7948 | 326 | int udf_get_filename(struct super_block *sb, uint8_t *sname, uint8_t *dname, |
cb00ea35 | 327 | int flen) |
1da177e4 LT |
328 | { |
329 | struct ustr filename, unifilename; | |
330 | int len; | |
331 | ||
4b11111a | 332 | if (udf_build_ustr_exact(&unifilename, sname, flen)) |
1da177e4 | 333 | return 0; |
1da177e4 | 334 | |
cb00ea35 CG |
335 | if (UDF_QUERY_FLAG(sb, UDF_FLAG_UTF8)) { |
336 | if (!udf_CS0toUTF8(&filename, &unifilename)) { | |
4b11111a MS |
337 | udf_debug("Failed in udf_get_filename: sname = %s\n", |
338 | sname); | |
1da177e4 LT |
339 | return 0; |
340 | } | |
cb00ea35 | 341 | } else if (UDF_QUERY_FLAG(sb, UDF_FLAG_NLS_MAP)) { |
4b11111a MS |
342 | if (!udf_CS0toNLS(UDF_SB(sb)->s_nls_map, &filename, |
343 | &unifilename)) { | |
344 | udf_debug("Failed in udf_get_filename: sname = %s\n", | |
345 | sname); | |
1da177e4 LT |
346 | return 0; |
347 | } | |
4b11111a | 348 | } else |
1da177e4 LT |
349 | return 0; |
350 | ||
28de7948 CG |
351 | len = udf_translate_to_linux(dname, filename.u_name, filename.u_len, |
352 | unifilename.u_name, unifilename.u_len); | |
4b11111a | 353 | if (len) |
1da177e4 | 354 | return len; |
28de7948 | 355 | |
1da177e4 LT |
356 | return 0; |
357 | } | |
358 | ||
28de7948 CG |
359 | int udf_put_filename(struct super_block *sb, const uint8_t *sname, |
360 | uint8_t *dname, int flen) | |
1da177e4 LT |
361 | { |
362 | struct ustr unifilename; | |
363 | int namelen; | |
364 | ||
4b11111a | 365 | if (!udf_char_to_ustr(&unifilename, sname, flen)) |
1da177e4 | 366 | return 0; |
1da177e4 | 367 | |
cb00ea35 | 368 | if (UDF_QUERY_FLAG(sb, UDF_FLAG_UTF8)) { |
28de7948 | 369 | namelen = udf_UTF8toCS0(dname, &unifilename, UDF_NAME_LEN); |
4b11111a | 370 | if (!namelen) |
1da177e4 | 371 | return 0; |
cb00ea35 | 372 | } else if (UDF_QUERY_FLAG(sb, UDF_FLAG_NLS_MAP)) { |
4b11111a MS |
373 | namelen = udf_NLStoCS0(UDF_SB(sb)->s_nls_map, dname, |
374 | &unifilename, UDF_NAME_LEN); | |
375 | if (!namelen) | |
1da177e4 | 376 | return 0; |
4b11111a | 377 | } else |
1da177e4 LT |
378 | return 0; |
379 | ||
380 | return namelen; | |
381 | } | |
382 | ||
383 | #define ILLEGAL_CHAR_MARK '_' | |
28de7948 CG |
384 | #define EXT_MARK '.' |
385 | #define CRC_MARK '#' | |
386 | #define EXT_SIZE 5 | |
1da177e4 | 387 | |
4b11111a MS |
388 | static int udf_translate_to_linux(uint8_t *newName, uint8_t *udfName, |
389 | int udfLen, uint8_t *fidName, | |
390 | int fidNameLen) | |
1da177e4 | 391 | { |
cb00ea35 | 392 | int index, newIndex = 0, needsCRC = 0; |
1da177e4 LT |
393 | int extIndex = 0, newExtIndex = 0, hasExt = 0; |
394 | unsigned short valueCRC; | |
395 | uint8_t curr; | |
396 | const uint8_t hexChar[] = "0123456789ABCDEF"; | |
397 | ||
28de7948 CG |
398 | if (udfName[0] == '.' && |
399 | (udfLen == 1 || (udfLen == 2 && udfName[1] == '.'))) { | |
1da177e4 LT |
400 | needsCRC = 1; |
401 | newIndex = udfLen; | |
402 | memcpy(newName, udfName, udfLen); | |
cb00ea35 CG |
403 | } else { |
404 | for (index = 0; index < udfLen; index++) { | |
1da177e4 | 405 | curr = udfName[index]; |
cb00ea35 | 406 | if (curr == '/' || curr == 0) { |
1da177e4 LT |
407 | needsCRC = 1; |
408 | curr = ILLEGAL_CHAR_MARK; | |
4b11111a MS |
409 | while (index + 1 < udfLen && |
410 | (udfName[index + 1] == '/' || | |
411 | udfName[index + 1] == 0)) | |
1da177e4 | 412 | index++; |
4b11111a MS |
413 | } |
414 | if (curr == EXT_MARK && | |
415 | (udfLen - index - 1) <= EXT_SIZE) { | |
416 | if (udfLen == index + 1) | |
1da177e4 | 417 | hasExt = 0; |
4b11111a | 418 | else { |
1da177e4 LT |
419 | hasExt = 1; |
420 | extIndex = index; | |
421 | newExtIndex = newIndex; | |
422 | } | |
423 | } | |
424 | if (newIndex < 256) | |
425 | newName[newIndex++] = curr; | |
426 | else | |
427 | needsCRC = 1; | |
428 | } | |
429 | } | |
cb00ea35 | 430 | if (needsCRC) { |
1da177e4 LT |
431 | uint8_t ext[EXT_SIZE]; |
432 | int localExtIndex = 0; | |
433 | ||
cb00ea35 | 434 | if (hasExt) { |
1da177e4 | 435 | int maxFilenameLen; |
4b11111a MS |
436 | for (index = 0; |
437 | index < EXT_SIZE && extIndex + index + 1 < udfLen; | |
438 | index++) { | |
1da177e4 LT |
439 | curr = udfName[extIndex + index + 1]; |
440 | ||
cb00ea35 | 441 | if (curr == '/' || curr == 0) { |
1da177e4 LT |
442 | needsCRC = 1; |
443 | curr = ILLEGAL_CHAR_MARK; | |
4b11111a MS |
444 | while (extIndex + index + 2 < udfLen && |
445 | (index + 1 < EXT_SIZE && | |
446 | (udfName[extIndex + index + 2] == '/' || | |
447 | udfName[extIndex + index + 2] == 0))) | |
1da177e4 LT |
448 | index++; |
449 | } | |
450 | ext[localExtIndex++] = curr; | |
451 | } | |
452 | maxFilenameLen = 250 - localExtIndex; | |
453 | if (newIndex > maxFilenameLen) | |
454 | newIndex = maxFilenameLen; | |
455 | else | |
456 | newIndex = newExtIndex; | |
4b11111a | 457 | } else if (newIndex > 250) |
1da177e4 LT |
458 | newIndex = 250; |
459 | newName[newIndex++] = CRC_MARK; | |
460 | valueCRC = udf_crc(fidName, fidNameLen, 0); | |
461 | newName[newIndex++] = hexChar[(valueCRC & 0xf000) >> 12]; | |
462 | newName[newIndex++] = hexChar[(valueCRC & 0x0f00) >> 8]; | |
463 | newName[newIndex++] = hexChar[(valueCRC & 0x00f0) >> 4]; | |
464 | newName[newIndex++] = hexChar[(valueCRC & 0x000f)]; | |
465 | ||
cb00ea35 | 466 | if (hasExt) { |
1da177e4 | 467 | newName[newIndex++] = EXT_MARK; |
cb00ea35 | 468 | for (index = 0; index < localExtIndex; index++) |
1da177e4 LT |
469 | newName[newIndex++] = ext[index]; |
470 | } | |
471 | } | |
28de7948 | 472 | |
1da177e4 LT |
473 | return newIndex; |
474 | } |